1 tbd Sun Microsystems, Inc.
4 2.6.16.60-0.23 (SLES 10),
5 2.6.18-92.1.6.el5 (RHEL 5),
6 2.6.22.14 vanilla (kernel.org).
7 * Client support for unpatched kernels:
8 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
9 2.6.16 - 2.6.21 vanilla (kernel.org)
10 * Recommended e2fsprogs version: 1.40.11-sun1
11 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
12 * RHEL 4 and RHEL 5/SLES 10 clients behaves differently on 'cd' to a
13 removed cwd "./" (refer to Bugzilla 14399).
17 Description: Properly propagate oinfo flags from lov to osc for statfs
18 Details : restore missing copy oi_flags to lov requests.
20 Severity : enhancement
22 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
23 : l_getgroups(8), lst(8), routerstat(8)
24 Details : included man pages for llobdstat(8), llstat(8),
25 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
27 Severity : enhancement
29 Description: configure's --enable-quota should check the
30 : kernel .config for CONFIG_QUOTA
31 Details : configure is terminated if --enable-quota is passed but
32 : no quota support is in kernel
36 Description: Remove portals compatibility
37 Details : Remove portals compatibility, not interoperable with releases
42 Description: Resolve device initialization race
43 Details : Prevent proc handler from accessing devices added to the
44 obd_devs array but yet be intialized.
46 Severity : enhancement
48 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
50 Severity : enhancement
52 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
57 Description: Using wrong pointer in osc_brw_prep_request
58 Details : Access to array[-1] can produce panic if kernel compiled with
59 CONFIG_PAGE_ALLOC enabled
63 Description: Client runs out of low memory
64 Details : Consider only lowmem when counting initial number of llap pages
68 Description: *optional* service tags registration
69 Details : if the "service tags" package is installed on a Lustre node
70 When the filesystem is mounted, a local-node service tag will
71 be created. See http://inventory.sun.com/ for more information
72 about the Service Tags asset management system.
76 Description: Kernel BUG tries to release flock
77 Details : Lustre does not destroy flock lock before last reference goes
78 away. So always drop flock locks when client is evicted and
79 perform unlock regardless of successfulness of speaking to MDS.
83 Description: add recount protection for osc callbacks, so avoid panic on shutdown
87 Description: sanity test 65a fails if stripecount of -1 is set
88 Details : handle -1 striping on filesystem in ll_dirstripe_verify
93 Description: ASSERTION(CheckWriteback(page,cmd)) failed
94 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
97 Severity : enhancement
99 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
103 Description: do not process already freed flock
104 Details : flock can possibly be freed by another thread before it reaches
105 to ldlm_flock_completion_ast.
109 Description: LBUG during stress test
110 Details : Need properly lock accesses the flock deadlock detection list.
114 Description: oops in page fault handler
115 Details : kernel page fault handler can return two special 'pages' in error case, don't
116 try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
120 Description: timeout with invalidate import.
121 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which should be
122 handled by ptlrpcd. This produce long age waiting and -ETIMEOUT
123 ptlrpc_invalidate_import and as result LASSERT.
125 Severity : enhancement
127 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
131 Description: enable MGS and MDT services start separately
132 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
133 a co-located MGS without starting the MGS, which is a complement
134 to 'nosvc' mount option.
138 Frequency : after recovery
139 Description: precreate to many object's after del orphan.
140 Details : del orphan st in oscc last_id == next_id and this triger growing
141 count of precreated objects. Set flag LOW to skip increase count
142 of precreated objects.
146 Frequency : rare, on clear nid stats
147 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
148 Details : when clean nid stats sometimes try destroy live entry,
149 and this produce panic in free.
153 Description: Stack overflow during MDS log replay
154 ease stack pressure by using a thread dealing llog_process.
158 Description: wait until IO finished before start new when do lock cancel.
159 Details : VM protocol want old IO finished before start new, in this case
160 need wait until PG_writeback is cleared until check dirty flag and
161 call writepages in lock cancel callback.
163 Severity : enhancement
165 Description: using special macro for print time and cleanup in includes.
169 Description: mds_mfd_close() ASSERTION(rc == 0)
170 Details : In mds_mfd_close(), we need protect inode's writecount change
171 within its orphan write semaphore to prevent possible races.
175 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
179 Frequency : rare, on shutdown ost
180 Description: don't hit live lock with umount ost.
181 Details : shrink_dcache_parent can be in long loop with destroy dentries,
182 use shrink_dcache_sb instead.
186 Description: don't panic with use echo client
187 Details : echo client pass NULL as client nid pointer and this produce null
192 Description: fix build on ppc32
193 Details : compile code with -m64 flag produce wrong object file for ppc32.
197 Description: add message levels for liblustreapi
201 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
202 Details : Move the 'good_osts' check before the 'total_bavail' check. This
203 will result in an -EAGAIN and in the exit call path we call
204 alloc_rr() which will with increasing aggressiveness attempt to
205 aquire precreated objects on the minimum number of required OSCs.
209 Description: Use old size assignment to avoid deadlock
210 Details : This reverts the changes in bugs 2369 and bug 14138 that introduced
211 the scheduling while holding a spinlock. We do not need locking
212 for size in ll_update_inode() because size is only updated from
213 the MDS for directories or files without objects, so there is no
214 other place to do the update, and concurrent access to such inodes
215 are protected by the inode lock.
219 Description: resolve "_IOWR redefined" build error on SLES10
223 Description: dump the memory debugging after all modules are unloaded to
224 suppress false negative in conf_sanity test 39
226 Severity : enhancement
228 Description: build kernel-ib packages for OFED 1.3 in our release cycle
233 Description: fix SLES kernel versioning
234 Details : the kernel version for our SLES 10 kernel did not include a "-"
235 before the "smp" at the end. while this was not a problem in
236 general, it did mean that software trying to use the kernel
237 version to try to detect a vendor specific kernel would fail.
238 this was most evident by the OFED build scripts.
242 Description: Don't update lov_desc members until making sure they are valid
243 Details : When updating lov_desc members via proc fs, need fix their
244 validities before doing the real update.
248 Description: don't put request into delay list while invalidate in flight.
249 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
250 import in flight. this produce timeout for invalidate and sometimes
251 can cause stale data.
255 Frequency : on ppc only
256 Description: not convert ost objects for directory because it's not exist.
257 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
261 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
263 Severity : enhancement
265 Description: Update to support 2.6.22.14 vanilla kernel.
269 Frequency : rare, on recovery
270 Description: read procfs can produce deadlock in some situation
271 Details : Holding lprocfs lock which send rpc can produce block for destroy
272 obd objects and this also block reconnect with -EALREADY. This isn't
273 fix all lprocfs bugs - but make it rare.
275 Severity : enhancement
277 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
280 Frequency : frequent on X2 node
282 Description: mdc_set_open_replay_data LBUG
283 Details : Set replay data for requests that are eligible for replay.
287 Description: lustre_mgs: operation 101 on unconnected MGS
288 Details : When MGC is disconnected from MGS long enough, MGS will evict the
289 MGC, and late on MGC cannot successfully connect to MGS and a lot
290 of the error messages complaining that MGS is not connected.
294 Frequency : on network error
295 Description: panic with double free request if network error
296 Details : mdc_finish_enqueue is finish request if any network error ocuring,
297 but it's true only for synchronus enqueue, for async enqueue
298 (via ptlrpcd) this incorrect and ptlrpcd want finish request
301 Severity : enhancement
303 Description: client-side metadata stat-ahead during readdir(directory readahead)
304 Details : perform client-side metadata stat-ahead when the client detects
305 readdir and sequential stat of dir entries therein
308 Frequency : on start mds
310 Description: Implement get_info(last_id) in obdfilter.
313 Frequency : occasional
315 Description: Correctly check stale fid, not start epoch if ost not support SOM
316 Details : open with flag O_CREATE need set old fid in op_fid3 because op_fid2
317 overwrited with new generated fid, but mds can anwer with one of these
318 two fids and both is not stale. setattr incorectly start epoch and
319 assume will be called done_writeting, but without SOM done_writing
323 Frequency : rare, depends on device drivers and load
325 Description: MDS or OSS nodes crash due to stack overflow
326 Details : Code changes in 1.8.0 increased the stack usage of some functions.
327 In some cases, in conjunction with device drivers that use a lot
328 of stack the MDS (or possibly OSS) service threads could overflow
329 the stack. One change which was identified to consume additional
330 stack has been reworked to avoid the extra stack usage.
333 Frequency : occasional
335 Description: Do not fail import if osc_interpret_create gets -EAGAIN
336 Details : If osc_interpret_create got -EAGAIN it immediately exits and
337 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
338 oscc_has_objects and see OSC has no objests and call
339 oscc_internal_create to resend create request.
341 Severity : enhancement
343 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
345 Severity : enhancement
347 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
350 Frequency : very rare
352 Description: Fix replay if there is an un-replied request and open
353 Details : In some cases, older replay request will revert the
354 mcd->mcd_last_xid on MDS which is used to record the client's
357 Severity : enhancement
359 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
361 Severity : enhancement
363 Description: Add rhel5 support to HEAD.
365 Serverity : enhancement
367 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
372 Description: Don't allow skipping OSTs if index has been specified.
373 Details : Don't allow skipping OSTs if index has been specified, make locking
374 in internal create lots better.
378 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
379 Details : access to bitfield in structure is always rounded to long
380 and this produce problem with not atomic change any bit.
384 Description: Lustre make rpms failed.
385 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
386 builds Lustre rpms from tarball.
389 Frequency : rare on shutdown ost
391 Description: If llog cancel was not send before clean_exports phase, this can
392 produce deadlock in llog code.
393 Details : If llog thread has last reference to obd and call class_import_put
394 this produce deadlock because llog_cleanup_commit_master wait when
395 last llog_commit_thread exited, but this never success because was
396 called from llog_commit_thread.
400 Description: allow userland application know is lost one of stripes.
401 Details : fill lvb_blocks with error code on ost and return it to
402 application if error flag found.
406 Description: NULL lov_tgts causing MDS oops
407 Details : more safe checks for NULL lov_tgts for avoid oops.
409 Severity : enhancement
411 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
415 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
416 Details : If an OST is down the MDS will hang indefinitely in
417 obd_statfs_rqset() waiting for the statfs data. While for
418 MDS QOS usage of statfs, it should not stuck in waiting.
420 Severity : enhancement
422 Description: remote_acl support
423 Details : Support ACL-based permission check for remote user.
424 Support setfacl/getfacl for remote user with the utils
425 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
426 the system "{s,g}etfacl" utils.
428 Severity : enhancement
430 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
432 Severity : enhancement
434 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
438 Description: make mgs_setparam() handle fsname containing dash
439 Details : fsname containing a dash does not work with lctl conf_param
441 Severity : enhancement
443 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
445 Severity : enhancement
447 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
451 Frequency : rare, at shutdown
452 Description: access already free / zero obd_namespace.
453 Details : if client_disconnect_export was called without force flag set,
454 and exist connect request in flight, this can produce access to
455 NULL pointer (or already free pointer) when connect_interpret
456 store ocd flags in obd_namespace.
460 Frequency : only at startup
461 Description: not alloc memory with spinlock held.
462 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
463 if any spinlock held.
465 Severity : enhancement
467 Description: make lustre randomly fail allocating memory
468 Details : Make lustre randomly failed allocating memory for testing purpose.
470 Severity : enhancement
472 Description: lost problems with lov objid file
473 Details : Fixes some scability and access to not inited memory problems
474 in work with lov objdid file.
479 Description: lfs find does not continue on file error
480 Details : Continue other files processing when a file/dir is absent.
484 Description: Inconsistent usage of lustre_pack_reply()
485 Details : Standardize the usage of lustre_pack_reply() such that it
486 always generate a CERROR on failure.
489 Frequency : occasional
491 Description: MDS hang or stay in waiting lock
492 Details : If client receive lock with CBPENDING flag ldlm need send lock
493 cancel as separate rpc, to avoid situation when cancel request
494 can't processed due all i/o threads stay in wait lock.
498 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
502 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
506 Description: Update to SLES9 kernel-2.6.5-7.287.3.
510 Description: Update to RHEL4 latest kernel.
512 Severity : enhancement
514 Description: Build SLES10 patchless client fails
515 Details : The configure was broken by run ./configure with
516 --with-linux-obj=.... argument for patchless client. When the
517 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
518 can't search header adequately. Use absolute path such as
519 -I($LINUX)/include instead.
523 Description: interrupt oig_wait produce painc on resend.
524 Details : brw_redo_request can be used for resend requests from ptlrpcd and
525 private set, and this produce situation when rq_ptlrpcd_data not
526 copyed to new allocated request and triggered LBUG on assert
527 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
528 ptlrpcd set if request is changed and can be safety changed to use
533 Description: LASSERT_{REQ,REP}SWAB macros are buggy
534 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
535 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
536 these macros and replace them with their difinitions instead.
540 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
541 Details : Update which_patch & target file for SLES10 latest kernel.
544 Description: Remove client patches from SLES 10 kernel.
545 Details : This causes SLES 10 clients to behave as patchless clients
546 even on a Lustre-patched (server) kernel.
548 Severity : enhancement
550 Description: self-adjustable client's lru lists
551 Details : use adaptive algorithm for managing client cached locks lru
552 lists according to current server load, other client's work
553 pattern, memory activities, etc. Both, server and client
554 side namespaces provide number of proc tunables for controlling
557 Severity : enhancement
559 Description: light-weight GSS support
560 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
561 and reduce performance overhead.
563 Severity : enhancement
565 Description: Linux keyring support
566 Details : Support using service of Linux keyring for Lustre GSS internal
567 context refresh/cache mechanism.
571 Description: Fix errors in lfs documentation
572 Details : Fixes man pages
576 Description: don't use GFP_* in generic Lustre code.
577 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
581 Description: obdclass is limited by single OBD_ALLOC(idarray)
582 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
586 Description: warnings with build patchless client with vanila 2.6.19 and up
587 Details : change old ctl_table style and replace ctl_table/ctl_table_header
588 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
592 Frequency : Only for SLES
593 Description: sanity_quota fail test_1
594 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
595 file, which makes TSTID[2] inunique.
599 Frequency : Only for SLES9
600 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
601 Details : Update target/ChangeLog/which_patch .
605 Frequency : Only for test_10 in sanity-quota.sh
606 Description: a bug in quota test code
607 Details : checking a test flag in a wrong place
611 Frequency : Only for test_13 in sanity-quota.sh
612 Description: a bug in quota test script
613 Details : shouldn't check overfull allocated quota
617 Description: jbd statistics
618 Details : Port older jbd statistics patch for sles10
620 Severity : enhancement
622 Description: Add printing inode info into message about error in writepage.
627 Description: reply_lock_interpret crash due to race with it and lock cancel.
628 Details : Do not replay locks that are being cancelled. Do not reference
629 locks by their address during replay, just by their handle.
633 Frequency : When flocks are used.
634 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
635 Details : Flock locks might destroy just granted lock if it could be merged
636 with another existing flock, this is done in completion handler,
637 so teach ldlm_cli_enquque_fini that this is a valid case for
643 Description: Oops in read and write path when failing to allocate lock.
644 Details : Check if lock allocation failed and return error back.
648 Description: lstripe command fails for valid OST index
649 Details : The stripe offset is compared to lov->desc.ld_tgt_count
650 instead of lov->desc.ld_active_tgt_count.
654 Description: sanity.sh failed test 103
655 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
660 Description: df doesn't work properly if diskfs blocksize != 4K
661 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
665 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
666 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
668 Severity : enhancement
670 Description: Adaptive timeouts
671 Details : RPC timeouts adapt to changing server load and network
672 conditions to reduce resend attempts and improve recovery time.
676 Description: llapi_file_create() does not allow some changes
677 Details : add llapi_file_open() that allows specifying the mode and
678 open flags, and also returns an open file handle.
682 Description: merge and cleanup kernel patches.
683 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
687 Description: Add journal checksum support.(Kernel part)
688 Details : The journal checksum feature adds two new flags i.e
689 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
690 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
691 indicates that the commit block contains the checksum for
692 the blocks described by the descriptor blocks. Now commit
693 record can be sent to disk without waiting for descriptor
694 blocks to be written to disk. This behavior is controlled
695 using JBD2_FEATURE_ASYNC_COMMIT flag.
699 Description: OSS needs mutliple precreate threads
700 Details : Add ability to start more than one create thread per OSS.
704 Description: Sanity.sh test_65k failed: setstripe should have succeeded
705 Details : execute lfs setstripe on client
709 Description: mds_obd_create error creating tmp object
710 Details : When the user sets quota on root, llog will be affected and can't
711 create files and write files.
714 Frequency : Always on ia64 patchless client, and possibly others.
716 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
717 Details : This allows the patchless client to be loaded on architectures
722 Description: RedHat Update kernel for RHEL5
723 Details : Add the kernel config file for RHEL5.
727 Description: RedHat Update kernel for RHEL5
728 Details : Modify the kernel config file more closer RHEL5.
732 Description: Build failure against Centos5 (RHEL5)
733 Details : Define PAGE_SIZE when it isn't present.
737 Description: test_7 on sanity-quota.sh can't be used on separate machines
738 Details : add facet to handle it
742 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
743 Details : with NFS, the anon dentry's parent was set to itself in
744 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
745 corresponding dentry other than use rec->ur_name.
747 Severity : enhancement
749 Description: lfs setstripe enhancement
750 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
754 Description: enable data checksumming by default
755 Details : enable checksum by default, allow --disable-checksum
756 configure option and "-o nochecksum" mount option. Checksums
757 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
761 Description: lustre support for RHEL5
762 Details : Add support for RHEL5.
766 Description: Client eviction due to failover config
767 Details : after a connection loss, the lustre client should attempt to
768 reconnect to the last active server first before trying the
769 other potential connections.
773 Description: when mds and osts use different quota unit(32bit and 64bit),
774 quota will be released repeatly.
775 Details : void sending multiple quota reqs to mds, which will keep the
776 status between the reqs.
780 Description: osts not allocated evenly to files
781 Details : change the condition to increase offset_idx
784 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
786 Description: Data corruption for OSTs that are formatted larger than 2TB
788 Details : When generating the bio request for lustre file writes the
789 sector number would overflow a temporary variable before being
790 used for the IO. The data reads correctly from Lustre (which
791 will overflow in a similar manner) but other file data or
792 filesystem metadata may be corrupted in some cases.
796 Description: Tune the kernel for good SCSI performance.
797 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
798 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
803 Description: rewrite ext2-derived code in obdclass/uuid.c
804 Details : rewrite inherited code (uuid parsing code from ext2 utils)
805 from scratch preserving functionality.
809 Description: "lfs find -obd UUID" prints directories
810 Details : "lfs find -obd UUID" will return all directory names instead
811 of just file names. It is incorrect because the directories
812 do not reside on the OSTs.
816 Description: use i_size_read and i_size_write in 2.6 port
817 Details : replace inode->i_size access with i_size_read/write()
820 Frequency : when removing large files
822 Description: scheduling issue during removal of large Lustre files
823 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
824 It causes scheduling issues when removing large files (17TB in the
828 Frequency : only with liblustre clients on XT3
830 Description: evictions taking too long
831 Details : allow llrd to evict clients directly on OSTs
834 Frequency : only on ppc
836 Description: /proc/fs/lustre/devices broken on ppc
837 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
838 We should make sure the type of 'index' is loff_t and then cast
839 explicitly as needed below. Do not assign an explicitly cast
843 Frequency : only for rhel5
845 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
846 Details : Modify the target file & which_kernel.
848 Severity : enhancement
850 Description: omit set fsid for export NFS
851 Details : fix set/restore device id for avoid EMFILE error and mark lustre fs
852 as FS_REQUIRES_DEV for avoid problems with generate fsid.
856 Frequency : Always, for kernels after 2.6.16
857 Description: Fix warning idr_remove called for id=.. which is not allocated.
858 Details : Last kernels save old s_dev before kill super and not allow
859 to restore from callback - restore it before call kill_anon_super.
863 Description: buffer overruns could theoretically occur
864 Details : llapi_semantic_traverse() modifies the "path" argument by
865 appending values to the end of the origin string, and a
866 overrun may occur. Adding buffer overrun check in liblustreapi.
870 Description: Fix error on 'ls .' at the top of the Lustre mount.
871 Details : Don't revalidate dentry if it is a root dentry.
875 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
876 Details : Modify vm-tunables-rhel4.patch.
880 Description: Kernel config for 2.6.18-vanilla.
881 Details : Modify targets/2.6-vanilla.target.in.
882 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
883 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
884 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
885 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
889 Description: improve handling recoverable errors
890 Details : if request processig with error which can be recoverable on server
891 request should be resend, otherwise page released from cache and
896 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
897 Details : Modify target file & which_patch.
898 A flaw was found in the IA32 system call emulation provided
899 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
900 value could be stored in the %RAX register, which could trigger an
901 out-of-bounds system call table access. An untrusted local user
902 could exploit this flaw to run code in the kernel
903 (ie a root privilege escalation). (CVE-2007-4573).
907 Description: O_DIRECT bypasses client statistics.
908 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
909 read_ahead_stats were updated. copied the stats section from
910 osc_send_oap_rpc() into async_internal().
914 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
918 Description: change order of libsysio includes
919 Details : '#include sysio.h' should always come before '#include xtio.h'
923 Description: handle "serious error: objid * already exists" more gracefully
924 Details : If LAST_ID value on disk is smaller than the objects existing in
925 the O/0/d* directories, it indicates disk corruption and causes an
926 LBUG(). If the object is 0-length, then we should use the existing
927 object. This will help to avoid a full fsck in most cases.
929 Severity : enhancement
931 Description: adapt the lustre_config script to support the upgrade case
932 Details : Add "-u" option for lustre_config script to support upgrading 1.4
933 server targets to 1.6 in parallel.
937 Description: To avoid grant space > avaible space when the disk is almost
938 full. Without this patch you might see the error "grant XXXX >
939 available" or some LBUG about grant, when the disk is almost
941 Details : In filter_check_grant, for non_grant cache write, we should
942 check the left space by if (*left > ungranted + bytes), instead
943 of (*left > ungranted), because only we are sure the left space
944 is enough for another "bytes", then the ungrant space should be
945 increase. In client, we should update cl_avail_grant only there
946 is OBD_MD_FLGRANT in the reply.
950 Description: Update RHEL 4 kernel to fix local root privilege escalation.
951 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
952 described in CVE-2007-4573. This problem could allow untrusted
953 local users to gain root access.
956 Frequency : when using O_DIRECT and quotas
958 Description: Incorrect file ownership on O_DIRECT output files
959 Details : block usage reported by 'lfs quota' does not take into account
960 files that have been written with O_DIRECT.
965 Description: touch file failed when fs is not full
966 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
967 otherwise we can get ENOSP while fs is not full.
971 Description: parallel lock callbacks
972 Details : Instead of sending blocking and completion callbacks as separated
973 requests, adding them to a set and sending in parallel.
976 Frequency : only for Cray XT3
977 Bugzilla : 12829/13455
978 Description: Changing primary group doesn't change the group lustre assigns to
980 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
981 overridden with the primary group provided by the group upcall,
982 whereas the supplied fsgid can be trusted if it is in the list of
983 supplementary groups returned by the group upcall.
985 Severity : enhancement
987 Description: Allow masking D_WARNING, D_ERROR messages from console
988 Details : Console messages can now be disabled via lnet.printk.
992 Description: User code with malformed file open parameter crashes client node
993 Details : Before packing join_file req, all the related reference should be
994 checked carefully in case some malformed flags cause fake
995 join_file req on client.
999 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
1000 Details : ldlm_completion_ast() assumes that a lock is granted when the req
1001 mode is equal to the granted mode. However, it should also check
1002 that LDLM_FL_CP_REQD is not set.
1006 Description: Heavy nfs access might result in deadlocks
1007 Details : After ELC code landed, it is now improper to enqueue any mds
1008 locks under och_sem, because enqueue might want to decide to
1009 cancel open locks for same inode we are holding och_sem for.
1013 Description: Client eviction while running blogbench
1014 Details : A lot of unlink operations with concurrent I/O can lead to a
1015 deadlock causing evictions. To address the problem, the number of
1016 oustanding OST_DESTROY requests is now throttled to
1017 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
1018 callbacks are processed in priority.
1022 Description: enable ACLs on MDS by default
1023 Details : ACLs must be enabled on MDS by default.
1026 Frequency : PPC/PPC64 only
1028 Description: conflicts between asm-ppc64/types.h and lustre_types.h
1029 Details : fix duplicated definitions between asm-ppc64/types.h and
1030 lustre_types.h on PPC.
1033 Frequency : PPC/PPC64 only
1035 Description: asm-ppc/segment.h does not exist
1036 Details : fix compile issue on PPC.
1040 Description: better handle error messages in extents code
1043 Frequency : RHEL4 only
1045 Description: mkfs is very slow on IA64/RHEL4
1046 Details : A performance regression has been discovered in the MPT Fusion
1047 driver between versions 3.02.73rh and 3.02.99.00rh. As a
1048 consequence, we have downgraded the MPT Fusion driver in the RHEL4
1049 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
1051 Severity : enhancement
1053 Description: SNMP support enhancement
1054 Details : Adding total number of sampled request for an MDS node in snmp
1057 Severity : enhancement
1059 Description: Optimize ldlm waiting list processing for PR extent locks
1060 Details : When processing waiting list for read extent lock and meeting read
1061 lock that is same or wider to it that is not contended, skip
1062 processing rest of the list and immediatelly return current
1063 status of conflictness, since we are guaranteed there are no
1064 conflicting locks in the rest of the list.
1068 Description: Time out and refuse to reconnect
1069 Details : When the failover node is the primary node, it is possible
1070 to have two identical connections in imp_conn_list. We must
1071 compare not conn's pointers but NIDs, otherwise we can defeat
1072 connection throttling.
1076 Description: port llog fixes from b1_6 into HEAD
1077 Details : Port llog reference couting and some llog cleanups from b1_6
1078 (bug 10800) into HEAD, for protect from panic and access to already
1079 free llog structures.
1083 Description: Detect stride IO mode in read-ahead
1084 Details : When a client does stride read, read-ahead should detect that and
1085 read-ahead pages according to the detected stride pattern.
1089 Description: data checksumming impacts single node performance
1090 Details : add support for several checksum algorithm. Currently, only CRC32
1091 and Adler-32 are supported. The checksum type can be changed on
1092 the fly via /proc/fs/lustre/osc/*/checksum_type.
1096 Description: use adler32 for page checksums
1097 Details : when available, use the Adler-32 algorithm instead of CRC32 for
1102 Description: build for x2 fails
1103 Details : fix compile issue on Cray systems.
1107 Description: Properly match for duplicate locks
1108 Details : Due to different lock order from skiplists code, we need to
1109 traverse entire list for now
1112 Frequency : only on PPC/SLES10
1114 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
1115 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
1116 be defined. Add a hack in mkfs_lustre.c to work around this compile
1121 Description: LASSERT on MDS when client holding flock lock dies
1122 Details : ldlm pool logic depends on number of granted locks equal to
1123 number of released locks which is not true for flock locks, so
1124 just exclude such locks from consideration.
1128 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
1129 Details : Use fsfilt_sync() for both the whole filesystem sync and
1130 individual file sync to eliminate dangerous inode locking
1131 with I_LOCK that can lead to a deadlock.
1135 Description: performance in 1.6.3
1136 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
1137 sure that 1MB requests are not fragmented by the block layer.
1139 Severity : enhancement
1141 Description: organize the server-side client stats on per-nid basis
1142 Details : Change the structure of stats under obdfilter and mds to
1150 The "uuid"s file would list the uuids of _active_ exports.
1151 And the clear entry is to clear all stats and stale nids.
1153 Severity : enhancement
1155 Description: eliminate client locks in face of contention
1156 Details : file contention detection and lockless i/o implementation
1157 for contended files.
1161 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
1164 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
1166 Description: MDS may lose file striping (and hence file data) in some cases
1167 Details : If there are additional extended attributes stored on the MDS,
1168 in particular ACLs, SELinux, or user attributes (if user_xattr
1169 is specified for the client mount options) then there is a risk
1170 of attribute loss. Additionally, the Lustre file striping
1171 needs to be larger than default (e.g. striped over all OSTs),
1172 and an additional attribute must be stored initially in the
1173 inode and then increase in size enough to be moved to the
1174 external attribute block (e.g. ACL growing in size) for file
1179 Description: skiplist implementation simplification
1180 Details : skiplists are used to group compatible locks on granted list
1181 that was implemented as tracking first and last lock of each lock group
1182 the patch changes that to using doubly linked lists
1186 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
1187 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
1188 because LASSERT on both the data supplied by a client, and the data
1189 on disk is dangerous and incorrect.
1191 Severity : enhancement
1193 Description: Slow truncate/writes to huge files at high offsets.
1194 Details : Directly associate cached pages to lock that protect those pages,
1195 this allows us to quickly find what pages to write and remove
1196 once lock callback is received.
1200 Description: more ldlm soft lockups
1201 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
1202 starve other threads from the resource lock for a long time in
1203 case of long waiting queue, so change the debug level from
1204 D_OTHER to the less frequently used D_INFO.
1206 Severity : enhancement
1208 Description: add -gid, -group, -uid, -user options to lfs find
1212 Description: Hung threads in invalidate_inode_pages2_range
1213 Details : The direct IO path doesn't call check_rpcs to submit a new RPC once
1214 one is completed. As a result, some RPCs are stuck in the queue
1219 Description: filter threads hungs on waiting journal commit
1220 Details : Cleanup filter group llog code, then only filter group llog will
1221 be only created in the MDS/OST syncing process.
1225 Description: Procfs and llog threads access destoryed import sometimes.
1226 Details : Sync the import destoryed process with procfs and llog threads by
1227 the import refcount and semaphore.
1229 Severity : enhancement
1231 Description: openlock cache of b1_6 port to HEAD
1236 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
1237 Details : If insertion of an extent fails, then discard the inode
1238 preallocation and free data blocks else it can lead to duplicate
1243 Description: don't always update ctime in ext3_xattr_set_handle()
1244 Details : Current xattr code updates the inode ctime in ext3_xattr_set_handle.
1245 In some cases the ctime should not be updated, for example for
1246 2.0->1.8 compatibility it is necessary to delete an xattr and it
1247 should not update the ctime.
1251 Bugzilla : 15713/16362
1252 Description: Assertion in iopen_connect_dentry in 1.6.3
1253 Details : looking up an inode via iopen with the wrong generation number can
1254 populate the dcache with a disconneced dentry while the inode
1255 number is in the process of being reallocated. This causes an
1256 assertion failure in iopen since the inode's dentry list contains
1257 both a connected and disconnected dentry.
1261 Description: assertion failure in ldlm_handle2lock()
1262 Details : fix a race between class_handle_unhash() and class_handle2object()
1263 introduced in lustre 1.6.5 by bug 13622.
1268 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
1269 Details : remove the limit of 256 scsi disks in the sd_iostat patch
1274 Description: oops in sd_iostats_seq_show()
1275 Details : unloading/reloading the scsi low level driver triggers a kernel
1276 bug when trying to access the sd iostat file.
1281 Description: Kernel panics during QLogic driver reload
1282 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
1283 patch, causing memory corruption.
1288 Description: journal_dev option does not work in b1_6
1289 Details : pass mount option during pre-mount.
1291 Severity : enhancement
1293 Description: Add a FIEMAP(FIle Extent MAP) ioctl
1294 Details : FIEMAP ioctl will allow an application to efficiently fetch the
1295 extent information of a file. It can be used to map logical blocks
1296 in a file to physical blocks in the block device.
1298 --------------------------------------------------------------------------------
1300 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
1302 * Support for kernels:
1303 2.6.5-7.283 (SLES 9),
1304 2.6.9-55.EL (RHEL 4),
1305 2.6.16.46-0.14 (SLES 10),
1306 2.6.18.8 vanilla (kernel.org)
1307 * Client support for unpatched kernels:
1308 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
1309 2.6.16 - 2.6.22 vanilla (kernel.org)
1310 * Due to recently discovered recovery problems, we do not recommend
1311 using patchless RHEL 4 clients with this or any earlier release.
1312 * Recommended e2fsprogs version: 1.39.cfs8
1313 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
1314 * Starting with this release, the ldiskfs backing filesystem required
1315 by Lustre is now in its own package, lustre-ldiskfs. This package
1316 should be installed. It is versioned separately from Lustre and
1317 may be released separately in future.
1321 Description: block reactivating mgc import until all deactivates complete
1322 Details : Fix race when failing back MDT/MGS to itself (testing)
1325 Frequency : at statup only
1327 Description: mds_lov_synchronize race leads to various problems
1328 Details : simultaneous MDT->OST connections at startup can cause the
1329 sync to abort, leaving the OSC in a bad state.
1331 Severity : enhancement
1333 Description: add optional extra BUILD_VERSION info
1334 Details : add a new environment variable (namely LUSTRE_VERS) which allows
1335 to override the lustre version.
1338 Frequency : 2.6.18 servers only
1340 Description: ll_kern_mount() doesn't release the module reference
1341 Details : The ldiskfs module reference count never drops down to 0
1342 because ll_kern_mount() doesn't release the module reference.
1347 Description: server LBUG when using old ost_num_threads parameter
1348 Details : Accept the old ost_num_threads parameter but warn that it
1349 is deprecated, and fix an off-by-one error that caused an LBUG.
1354 Description: Transient SCSI error results in persistent IO issue
1355 Details : iobuf->dr_error is not reinitialized to 0 between two
1359 Frequency : sometimes when underlying device returns I/O errors
1361 Description: OSTs not going read-only during write failures
1362 Details : OSTs are not remounted read-only when the journal commit threads
1363 get I/O errors because fsfilt_ext3 calls journal_start/stop()
1364 instead of the ext3 wrappers.
1368 Description: poor connect scaling with increasing client count
1369 Details : Don't run filter_grant_sanity_check for more than 100 exports
1370 to improve scaling for large numbers of clients.
1373 Frequency : SLES10 only
1375 Description: sanity-quota.sh quotacheck failed: rc = -22
1376 Details : Quotas cannot be enabled on SLES10.
1379 Frequency : liblustre clients only
1381 Description: getdirentries does not give error when run on compute nodes
1382 Details : getdirentries does not fail when the size specified as an argument
1383 is too small to contain at least one entry
1385 Severity : enhancement
1387 Description: Add LNET router traceability for debug purposes
1388 Details : If a checksum failure occurs with a router as part of the
1389 IO path, the NID of the last router that forwarded the bulk data
1390 is printed so it can be identified.
1395 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
1396 Details : Due to a race condition, liblustre clients were occasionally
1397 evicted incorrectly.
1399 Severity : enhancement
1401 Description: lfs setstripe use optional parameters instead of postional
1404 Severity : enhancement
1406 Description: Nanosecond timestamp support for ldiskfs
1407 Details : The on-disk ldiskfs filesystem has added support for nanosecond
1408 resolution timestamps. There is not yet support for this at
1409 the Lustre filesystem level.
1412 Frequency : during server recovery
1414 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
1415 Details : request with rq_no_resend flag not awake l_wait_event if they get a
1419 Frequency : nfs export on patchless client
1421 Description: connectathon hang when test nfs export over patchless client
1422 Details : Disconnected dentry cannot be found with lookup, so we do not need
1423 to unhash it or make it invalid
1426 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
1427 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
1428 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
1433 Description: open req refcounting wrong on reconnect
1434 Details : If reconnect happened between getting open reply from server and
1435 call to mdc_set_replay_data in ll_file_open, we will schedule
1436 replay for unreferenced request that we are about to free.
1437 Subsequent close will crash in variety of ways.
1438 Check that request is still eligible for replay in
1439 mdc_set_replay_data().
1444 Description: disable writes to filesystem when reading health_check file
1445 Details : the default for reading the health_check proc file has changed
1446 to NOT do a journal transaction and write to disk, because this
1447 can cause reads of the /proc file to hang and block HA state
1448 checking on a healthy but otherwise heavily loaded system. It
1449 is possible to return to the previous behaviour during configure
1450 with --enable-health-write.
1452 Severity : enhancement
1454 Description: 64-bit inode version
1455 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
1456 to the inode. This will be required for version-based recovery.
1461 Description: MDS fails to start if a duplicate client export is detected
1462 Details : in some rare cases it was possible for a client to connect to
1463 an MDS multiple times. Upon recovery the MDS would detect this
1464 and fail during startup. Handle this more gracefully.
1466 Severity : enhancement
1468 Description: Add -o localflock option to simulate old noflock
1470 Details : This will achieve local-only flock/fcntl locks
1476 Description: log_commit_thread vs filter_destroy race leads to crash
1477 Details : Take import reference before releasing llog record semaphore
1482 Description: Wrong request locking in request set processing
1483 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
1484 imp_delayed_list, in this place should be used imp_lock.
1487 Frequency : when reconnection
1489 Description: Grant Leak when osc reconnect to OST
1490 Details : When osc reconnect ost, OST(filter) should check whether it
1491 should grant more space to client by comparing fed_grant and
1492 cl_avail_grant, and return the granted space to client instead
1493 of "new granted" space, because client will call osc_init_grant
1494 to update the client grant space info.
1497 Frequency : when client reconnect to OST
1499 Description: Grant Leak when osc do resend and replay bulk write
1500 Details : When osc reconnect to OST, OST(filter)should clear grant info of
1501 bulk write request, because the grant info while be sync between
1502 OSC and OST when reconnect, and we should ignore the grant info
1503 these of resend/replay write req.
1508 Description: Grant space more than avaiable left space sometimes.
1509 Details : When then OST is about to be full, if two bulk writing from
1510 different clients came to OST. Accord the avaliable space of the
1511 OST, the first req should be permitted, and the second one
1512 should be denied by ENOSPC. But if the seconde arrived before
1513 the first one is commited. The OST might wrongly permit second
1514 writing, which will cause grant space > avaiable space.
1517 Frequency : when client is evicted
1519 Description: Grant might be wrongly erased when osc is evicted by OST
1520 Details : when the import is evicted by server, it will fork another
1521 thread ptlrpc_invalidate_import_thread to invalidate the
1522 import, where the grant will be set to 0. While the original
1523 thread will update the grant it got when connecting. So if
1524 the former happened latter, the grant will be wrongly errased
1525 because of this race.
1530 Description: Checking Stale with correct fid
1531 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
1532 to check whether it is stale, because sometimes, we want the
1533 enqueue happened anyway, and op_data.fid2 will not be initialized.
1535 Severity : enhancement
1537 Description: update patchless client
1538 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
1541 Frequency : only with 2.4 kernel
1543 Description: random memory corruption
1544 Details : size of struct ll_inode_info is to big for union inode.u and this
1545 can be cause of random memory corruption.
1550 Description: Memory leak in recovery
1551 Details : Lov_mds_md was not free in an error handler in mds_create_object.
1552 It should also check obd_fail before fsfilt_start, otherwise if
1553 fsfilt_start return -EROFS,(failover mds during mds recovery).
1554 then the req will return with repmsg->transno = 0 and rc = EROFS.
1555 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
1556 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
1557 be freed no matter whether fsfilt_commit success or not.
1560 Frequency : only with huge count clients
1562 Description: Prevents from taking the superblock lock in llap_from_page for
1564 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
1565 for llap_from_page call in ll_removepage prevents from taking the
1566 superblock lock for a soon died page.
1571 Description: Not check open intent error before release open handle
1572 Details : in some rare cases, the open intent error is not checked before
1573 release open handle, which may cause
1574 ASSERTION(open_req->rq_transno != 0), because it tries to release
1575 the failed open handle.
1580 Description: Set cat log bitmap only after create log success.
1581 Details : in some rare cases, the cat log bitmap is set too early. and it
1582 should be set only after create log success.
1586 Description: Accessing a block bevice can re-enable I/O when Lustre is
1587 tearing down a device.
1588 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
1593 Description: service threads may hog cpus when there are a lot of requests
1594 Details : Insert cond_resched to give other threads a chance to use some CPU
1599 Description: the cat log was not initialized in recovery
1600 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
1601 unlink log on mds will not be initialized until mds post
1602 recovery. And also in mds post recovery, the unlink log will
1603 initialization will be done asynchronausly, so there will be race
1604 between add unlink log and unlink log initialization.
1608 Description: brw_stats were being printed incorrectly
1609 Details : brw_stats were being printed as log2 but all of them were not
1610 recorded as log2. Also remove some code duplication arising from
1611 filter_tally_{read,write}.
1615 Frequency : rare, only in recovery.
1616 Description: ASSERTION(req->rq_type != LI_POISON) failed
1617 Details : imp_lock should be held while iterating over imp_sending_list for
1618 prevent destroy request after get timeout in ptlrpc_queue_wait.
1622 Description: replay-single.sh test 52 fails
1623 Details : A lock's skiplist need to be cleanup when it being unlinked
1624 from its resource list.
1628 Description: Short directio read returns full requested size rather than
1630 Details : Direct I/O operations should return actual amount of bytes
1631 transferred rather than requested size.
1633 Severity : enhancement
1635 Description: metadata RPC reduction (e.g. for rm performance)
1636 Details : decrease the amount of synchronous RPC between clients and servers
1637 by canceling conflicing lock before the operation on the client
1638 and packing thier handles into the main operation RPC to server.
1640 Severity : enhancement
1642 Description: Async OSC create to avoid the blocking unnecessarily.
1643 Details : If a OST has no remain object, system will block on the creating
1644 when need to create a new object on this OST. Now, ways use
1645 pre-created objects when available, instead of blocking on an
1646 empty osc while others are not empty. If we must block, we block
1647 for the shortest possible period of time.
1649 Severity : enhancement
1651 Description: refine locking for avoid write wrong info into lov_objid file
1652 Details : fix possible races with add new target and write/update data in
1655 --------------------------------------------------------------------------------
1657 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
1662 Frequency : on some architectures
1664 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
1665 Details : Uninitialized flags sometimes cause configuration commands to
1669 Frequency : patchless clients only
1671 Description: missing __iget() symbol export
1672 Details : The __iget() symbol export is missing. To avoid the need for
1673 this on patchless clients the deathrow inode reaper is turned
1674 off, and we depend on the VM to clean up old inodes. This
1675 dependency was during via the fix for bug 12181.
1677 --------------------------------------------------------------------------------
1679 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
1681 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
1682 INTEROPERATE with older versions automatically. In many cases a
1683 special upgrade step is needed. Please read the
1684 user documentation before upgrading any part of a live system.
1685 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
1686 version will not interoperate with 1.6 betas before beta5 (1.5.95).
1687 * WARNING: Lustre configuration and startup changes are required with
1688 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
1694 Severity : enhancement
1696 Description: MountConf
1697 Details : Lustre configuration is now managed via mkfs and mount
1698 commands instead of lmc and lconf. New obd types (MGS, MGC)
1699 are added for dynamic configuration management. See
1700 https://mail.clusterfs.com/wikis/lustre/MountConf for
1703 Severity : enhancement
1705 Description: dynamic OST addition
1706 Details : OSTs can now be added to a live filesystem
1708 Severity : enhancement
1710 Description: startup order invariance
1711 Details : MDTs and OSTs can be started in any order. Clients only
1712 require the MDT to complete startup.
1714 Severity : enhancement
1716 Description: parallel, asynchronous orphan cleanup
1717 Details : orphan cleanup is now performed in separate threads for each
1718 OST, allowing parallel non-blocking operation.
1720 Severity : enhancement
1722 Description: optimized stripe assignment
1723 Details : stripe assignments are now made based on ost space available,
1724 ost previous usage, and OSS previous usage, in order to try
1725 to optimize storage space and networking resources.
1727 Severity : enhancement
1729 Description: Permanently set tunables
1730 Details : All writable /proc/fs/lustre tunables can now be permanently
1731 set on a per-server basis, at mkfs time or on a live system.
1733 Severity : enhancement
1735 Description: Lustre message v2
1736 Details : Add lustre message format v2.
1738 Severity : enhancement
1740 Description: client OST exclusion list
1741 Details : Clients can be started with a list of OSTs that should be
1742 declared "inactive" for known non-responsive OSTs.
1746 Description: SPEC SFS validation failure on NFS v2 over lustre.
1747 Details : Changes the blocksize for regular files to be 2x RPC size,
1748 and not depend on stripe size.
1750 Severity : enhancement
1752 Description: Multiple MD RPCs in flight.
1753 Details : Further unserialise some read-only MDS RPCs - learn about intents.
1754 To avoid overly-overloading MDS, introduce a limit on number of
1755 MDS RPCs in flight for a single client and add /proc controls
1756 to adjust this limit.
1758 Severity : enhancement
1760 Description: client read/write statistics
1761 Details : Add client read/write call usage stats for performance
1762 analysis of user processes.
1763 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
1764 file access. extents_stats shows chunk size distribution.
1765 extents_stats_per_process show chunk size distribution per
1768 Severity : enhancement
1770 Description: mds statistics
1771 Details : Add detailed mds operations statistics in
1772 /proc/fs/lustre/mds/*/stats.
1776 Description: Failure of copying files with lustre special EAs.
1777 Details : Client side always return success for setxattr call for lustre
1778 special xattr (currently only "trusted.lov").
1783 Description: Refcount LNET uuids
1784 Details : The global LNET uuid list grew linearly with every startup;
1785 refcount repeated list entries instead of always adding to
1788 Severity : enhancement
1790 Description: Dynamic service threads
1791 Details : Within a small range, start extra service threads
1792 automatically when the request queue builds up.
1795 Frequency : mixed-endian client/server environments
1797 Description: mixed-endian crashes
1798 Details : The new msg_v2 system had some failures in mixed-endian
1801 Severity : enhancement
1803 Description: Easy OST removal
1804 Details : OSTs can be permanently deactivated with e.g. 'lctl
1805 conf_param lustre-OST0001.osc.active=0'
1807 Severity : enhancement
1809 Description: MGS proc entries
1810 Details : Added basic proc entries for the MGS showing what filesystems
1813 Severity : enhancement
1815 Description: provide MGS failover
1816 Details : Added config lock reacquisition after MGS server failover.
1818 Severity : enhancement
1820 Description: add Linux 2.4 support
1821 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
1825 Description: a large application tries to do I/O to the same resource and dies
1826 in the middle of it.
1827 Details : Check the req->rq_arrival time after the call to
1828 ost_brw_lock_get(), but before we do anything about
1829 processing it & sending the BULK transfer request. This
1830 should help move old stale pending locks off the queue as
1831 quickly as obd_timeout.
1834 Frequency : when an incorrect nid is specified during startup
1836 Description: ptlrpc connect to non-existant node causes kernel crash
1837 Details : LNET can't be re-entered from an event callback, which
1838 happened when we expire a message after the export has been
1839 cleaned up. Instead, hand the zombie cleanup off to another
1842 Severity : enhancement
1844 Description: plain/inodebits lock performance improvement
1845 Details : Grouping plain/inodebits in granted list by their request modes
1846 and bits policy, thus improving the performance of search through
1850 Frequency : only if OST filesystem is corrupted
1852 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1853 Details : for a short time RPCs with bulk IO are in the replay list,
1854 but replay of bulk IOs is unimplemented. If the OST filesystem
1855 is corrupted due to disk cache incoherency and then replay is
1856 started it is possible to trip an assertion. Avoid putting
1857 committed RPCs into the replay list at all to avoid this issue.
1860 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1862 Description: System hang on startup
1863 Details : This bug allowed the liblustre (e.g. catamount) client to
1864 return to the app before handling all startup RPCs. This
1865 could leave the node unresponsive to lustre network traffic
1866 and manifested as a server ptllnd timeout.
1868 Severity : enhancement
1870 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1871 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
1872 boolean to control whether to print peer debug info when a
1873 client's RPC times out.
1876 Frequency : only for kernels with patches from Lustre below 1.4.3
1878 Description: Remove old rdonly API
1879 Details : Remove old rdonly API which unused from at least lustre 1.4.3
1882 Frequency : only for devices with external journals
1884 Description: Set external device read-only also
1885 Details : During a commanded failover stop, we set the disk device
1886 read-only while the server shuts down. We now also set any
1887 external journal device read-only at the same time.
1890 Frequency : when upgrading from 1.4 while trying to change parameters
1892 Description: The wrong (new) MDC name was used when setting parameters for
1893 upgraded MDT's. Also allows changing of OSC (and MDC)
1894 parameters if --writeconf is specified at tunefs upgrade time.
1897 Frequency : when setting specific ost indicies
1899 Description: QOS code breaks on skipped indicies
1900 Details : Add checks for missing OST indicies in the QOS code, so OSTs
1901 created with --index need not be sequential.
1903 Severity : enhancement
1905 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1906 Details : The uninit_groups feature works in conjunction with the kernel
1907 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1908 up the pass1 processing of e2fsck. This is a read-only feature
1909 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1910 mount filesystems that have had this feature enabled.
1912 Severity : enhancement
1914 Description: Improve multi-block allocation algorithm to avoid fragmentation
1915 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1916 allocation locality and avoid filesystem fragmentation.
1918 ------------------------------------------------------------------------------
1920 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
1922 * Support for kernels:
1923 2.4.21-47.0.1.EL (RHEL 3)
1924 2.6.5-7.283 (SLES 9)
1925 2.6.9-42.0.10.EL (RHEL 4)
1926 2.6.12.6 vanilla (kernel.org)
1927 2.6.16.27-0.9 (SLES 10)
1928 * Recommended e2fsprogs version: 1.39.cfs5
1930 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1934 Frequency : occasional, depends on client load and configuration
1935 Bugzilla : 12181, 12203
1936 Description: data loss for recently-modified files
1938 Details : In some cases it is possible that recently written or created
1939 files may not be written to disk in a timely manner (this should
1940 normally be within 30s unless client IO load is very high).
1941 The problem appears as zero-length files or files that are a
1942 multiple of 1MB in size after a client crash or client eviction
1943 that are missing data at the end of the file.
1945 This problem is more likely to be hit on clients where files are
1946 repeatedly created and unlinked in the same directory, clients
1947 have a large amount of RAM, have many CPUs, the filesystem has
1948 many OSTs, the clients are rebooted frequently, and/or the files
1949 are not accessed by other nodes after being written.
1951 The presence of the problem can be detected by looking at
1952 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
1953 smaller than the second (nr_unused) then dirty files will not
1954 be flushed automatically to disk. "sync; sleep 10" should be
1955 run several times on the node before unmounting it to update
1956 Lustre (this is also safe to run on nodes without this problem).
1958 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
1959 that can cause this same problem, so customers using that kernel
1960 also need to update the kernel in addition to Lustre. In order
1961 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
1963 It is normal that files written just before a client crash (less
1964 than 30s) may not yet have been flushed to disk, even for local
1968 Frequency : frequent on thin XT3 nodes
1970 Description: UUID collision on thin XT3 Linux nodes
1971 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
1972 randomly, since we relied on an insufficiently-seeded PRNG.
1977 Description: OSS hangs after "All ost request buffers busy"
1978 Details : A deadlock between quota and journal operations caused OSS
1979 hangs after printing "All ost request buffers busy."
1982 Frequency : always on liblustre builds
1984 Description: Cleanup compiler warnings on liblustre
1987 Frequency : always on liblustre builds on XT3
1989 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
1993 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
1994 > 32000 subdirectories
1995 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
1996 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
1997 subdirectory count crosses 32000. This will aid e2fsck to
1998 correctly handle more than 32000 subdirectories.
2001 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
2003 Description: System hang on startup
2004 Details : This bug allowed the liblustre (e.g. catamount) client to
2005 return to the app before handling all startup RPCs. This
2006 could leave the node unresponsive to lustre network traffic
2007 and manifested as a server ptllnd timeout.
2009 Severity : enhancement
2011 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
2012 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
2013 boolean to control whether to print peer debug info when a
2014 client's RPC times out.
2019 Description: make O_SYNC working on 2.6 kernels
2020 Details : 2.6 kernels use different method for mark pages for write,
2021 so need add a code to lustre for O_SYNC work.
2026 Description: Failure to close file and release space on NFS
2027 Details : Put inode details into lock acquired in ll_intent_file_open.
2028 Use mdc_intent_lock in ll_intent_open to properly
2029 detect all kind of errors unhandled by mdc_enqueue.
2034 Description: proc file read during shutdown sometimes raced obd removal,
2036 Details : Add lock to prevent obd access after proc file removal.
2039 Frequency : Only for files larger than 4GB on 32-bit clients.
2041 Description: improperly doing page alignment of locks
2042 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
2043 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
2048 Description: RPCs being resent when they shouldn't be
2049 Details : Some RPCs that should not be resent are being resent. This
2050 can cause inconsistencies in the RPC state machine. Do not
2051 resend such requests.
2054 Frequency : rare, only with NFS export
2056 Description: Crash on NFS re-export node
2057 Details : under very unusual load conditions an assertion is hit in
2058 ll_intent_file_open()
2061 Frequency : only if OST filesystem is corrupted
2063 Description: client incorrectly hits assertion in ptlrpc_replay_req()
2064 Details : for a short time RPCs with bulk IO are in the replay list,
2065 but replay of bulk IOs is unimplemented. If the OST filesystem
2066 is corrupted due to disk cache incoherency and then replay is
2067 started it is possible to trip an assertion. Avoid putting
2068 committed RPCs into the replay list at all to avoid this issue.
2073 Description: large O_DIRECT requests fail under memory pressure/fragmentation
2074 Details : Large single O_DIRECT read and write calls can fail to allocate
2075 a sufficiently large buffer to process the request. In case of
2076 allocation failure the allocation is retried with a smaller
2077 buffer and broken into smaller requests.
2079 Severity : enhancement
2081 Description: Add -o localflock option to simulate old noflock behaviour.
2082 Details : This will achieve local-only flock/fcntl locks coherentness.
2087 Description: versioning check is incomplete
2088 Details : Checking the version difference of client vs. server, report
2089 error if the gap is too big.
2094 Description: add support PG_writeback bit
2095 Details : add support for PG_writeback bit for Lustre, for more carefull
2096 work with page cache in 2.6 kernel. This also fix some deadlocks
2097 and remove hack for work O_SYNC with 2.6 kernel.
2099 Severity : enhancement
2101 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
2102 Details : The uninit_groups feature works in conjunction with the kernel
2103 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
2104 up the pass1 processing of e2fsck. This is a read-only feature
2105 in ldiskfs2 only, so older kernels and current ldiskfs cannot
2106 mount filesystems that have had this feature enabled.
2108 Severity : enhancement
2110 Description: Improve multi-block allocation algorithm to avoid fragmentation
2111 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
2112 allocation locality and avoid filesystem fragmentation.
2114 ------------------------------------------------------------------------------
2116 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
2118 * Support for kernels:
2119 2.6.9-42.0.3.EL (RHEL 4)
2120 2.6.5-7.276 (SLES 9)
2121 2.4.21-47.0.1.EL (RHEL 3)
2122 2.6.12.6 vanilla (kernel.org)
2123 2.6.16.21-0.8 (SLES10)
2124 * Recommended e2fsprogs version: 1.39.cfs2-0
2126 * The backwards-compatible /proc/sys/portals symlink has been removed
2127 in this release. Before upgrading, please ensure that you change
2128 any configuration scripts or /etc/sysctl.conf files that access
2129 /proc/sys/portals/* or sysctl portals.* to use the corresponding
2130 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
2131 in advance of the upgrade on any system running Lustre 1.4.6 or
2132 newer, since /proc/sys/lnet was added in that version.
2133 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
2137 Frequency : only when quota is used
2139 Description: avoid scanning export list for quota master
2140 Details : Change the algorithms to avoid scanning export list in order
2141 to improve the efficiency.
2144 Frequency : MDS failover only, very rarely
2146 Description: "went back in time" messages on mds failover
2147 Details : The greatest transno may be lost when the current operation
2148 finishes with an error (transno==0) and the client's last_rcvd
2149 record is over-written. Save the greatest transno in the
2150 mds_last_transno for this case.
2153 Frequency : always for specific kernels and striping counts
2155 Description: client may get "Matching packet too big" without ACL support
2156 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
2157 when trying to access files in certain configurations. The
2158 clients should in fact be denied when mounting because they do
2159 not understand ACLs.
2162 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
2164 Description: many clients connecting with IO in progress causes connect timeouts
2165 Details : Avoid synchronous journal commits to avoid delays caused by many
2166 clients connecting/disconnecting when bulk IO is in progress.
2167 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
2168 OST_IO_PORTAL to avoid delays behind potentially many pending
2172 Frequency : occasionally with multiple writers to a single file
2174 Description: shared writes to file may result in wrong size reported by stat()
2175 Details : Allow growing of kms when extent lock is cancelled
2178 Frequency : always with random mmap IO to multi-striped file
2180 Description: mmap write might be lost if we are writing to a 'hole' in stripe
2181 Details : Only if the hole is at the end of OST object so that kms is too
2182 small. Fix is to increase kms accordingly in ll_nopage.
2185 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
2187 Description: writes to a missing object would leak memory on the OST
2188 Details : If there is an inconsistency between the MDS and OST filesystems,
2189 such that the MDS references an object that doesn't exist, writes
2190 to that object will leak memory due to incorrect cleanup in the
2191 error handling path, eventually running out of memory on the OST.
2196 Description: Creating too long symlink causes lustre errors
2197 Details : Check symlink and name lengths before sending requests to MDS.
2200 Frequency : only if flock is enabled (not on by default)
2202 Description: posix locks not released on fd closure on 2.6.9+
2203 Details : We failed to add posix locks to list of inode locks on 2.6.9+
2204 kernels, this caused such locks not to be released on fd close and
2205 then assertions on fs unmount about still used locks.
2208 Frequency : MDS failover only, very rarely
2210 Description: clients may get ASSERTION(granted_lock != NULL)
2211 Details : When request was taking a long time, and a client was resending
2212 a getattr by name lock request. The were multiple lock requests
2213 with the same client lock handle and
2214 mds_getattr_name->fixup_handle_for_resent_request found one of the
2215 lock handles but later failed with ASSERTION(granted_lock != NULL).
2220 Description: handle->h_buffer_credits > 0, assertion failure
2221 Details : h_buffer_credits is zero after truncate, causing assertion
2222 failure. This patch extends the transaction or creates a new
2226 Frequency : NFS re-export or patchless client
2227 Bugzilla : 11179, 10796
2228 Description: Crash on NFS re-export node (__d_move)
2229 Details : We do not want to hash the dentry if we don't have a lock.
2230 But if this dentry is later used in d_move, we'd hit uninitialised
2231 list head d_hash, so we just do this to init d_hash field but
2232 leave dentry unhashed.
2235 Frequency : NFS re-export or patchless client
2237 Description: NFS exports has problem with symbolic link
2238 Details : lustre client didn't properly install dentry when re-exported
2239 to NFS or running patchless client.
2242 Frequency : NFS re-export or patchless client
2244 Description: Various nfs/patchless fixes.
2245 Details : fixes reuse disconected alias for lookup process - this fixes
2246 warning "find_exported_dentry: npd != pd",
2247 fix permission error with open files at nfs.
2248 fix apply umask when do revalidate.
2251 Frequency : occasional
2253 Description: Crash on NFS re-export node
2254 Details : calling clear_page() on the wrong pointer triggered oops in
2255 generic_mapping_read().
2258 Frequency : rarely, using O_DIRECT IO
2260 Description: unaligned directio crashes client with LASSERT
2261 Details : check for unaligned buffers before trying any requests.
2264 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
2266 Description: stale data returned from RAID cache
2267 Details : If only a small amount of IO is done to the RAID device before
2268 reading it again it is possible to get stale data from the RAID
2269 cache instead of reading it from disk.
2272 Frequency : always for sles10 kernel
2274 Description: sles10 support
2275 Details : ll_follow_link: compile fixes and using of nd_set_link
2276 under newer kernels.
2279 Frequency : depends on arch, kernel and compiler version, always on sles10
2282 Description: recursive or deep enough symlinks cause stack overflow
2283 Details : getting rid of large stack-allocated variable in
2287 Frequency : depends on hardware
2289 Description: lustre write performance loss in the SLES10 kernel
2290 Details : the performance loss is caused by using of write barriers in the
2291 ext3 code. The SLES10 kernel turns barrier support on by
2292 default. The fix is to undo that change for ldiskfs.
2294 ------------------------------------------------------------------------------
2296 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
2298 * Support for kernels:
2299 2.6.9-42.0.3EL (RHEL 4)
2300 2.6.5-7.276 (SLES 9)
2301 2.4.21-47.0.1.EL (RHEL 3)
2302 2.6.12.6 vanilla (kernel.org)
2306 Frequency : quota enabled and large files being deleted
2308 Description: releasing more than 4GB of quota at once hangs OST
2309 Details : If a user deletes more than 4GB of files on a single OST it
2310 will cause the OST to spin in an infinite loop. Release
2311 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2316 Description: statfs data retrieved from /proc may be stale or zero
2317 Details : When reading per-device statfs data from /proc, in the
2318 {kbytes,files}_{total,free,avail} files, it may appear
2319 as zero or be out of date.
2322 Frequency : systems with MD RAID1 external journal devices
2324 Description: lconf's call to blkid is confused by RAID1 journal devices
2325 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2326 of returning all block devices that match the journal UUID.
2329 Frequency : always, for aggregate stripe size over 4GB
2331 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
2332 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
2333 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
2336 Frequency : always if "lfs find" used on a local file/directory
2338 Description: "lfs find" segfaults if used on a local file/directory
2339 Details : The case where a directory component was not specified wasn't
2340 handled correctly. Handle this properly.
2343 Frequency : always on ppc64
2345 Description: the write to an ext3 filesystem mounted with mballoc got stuck
2346 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2347 perform endianness conversion.
2350 Frequency : rarely (truncate to non-zero file size after write under load)
2351 Bugzilla : 10730, 10687
2352 Description: Files padded with zeros to next 4K multiple
2353 Details : With filesystems mounted using the "extents" option (2.6 kernels)
2354 it is possible that files that are truncated to a non-zero size
2355 immediately after being written are filled with zero bytes beyond
2356 the truncated size. No file data is lost.
2358 Severity : enhancement
2360 Description: Allow recovery/failover for liblustre clients.
2361 Details : liblustre clients were unaware of failover configurations until
2364 Severity : enhancement
2366 Description: user file locks should fail when not mounting with flock option
2367 Details : Set up an error-returning stub in ll_file_operations.lock field
2368 to prevent incorrect behaviour when client is mounted without
2369 flock option. Also, set up properly f_op->flock field for
2373 Frequency : always on ia64
2375 Description: "lfs df" loops on printing out MDS statfs information
2376 Details : The obd_ioctl_data was not initialized and in some systems
2377 this caused a failure during the ioctl that did not return
2378 an error. Initialize the struct and return an error on failure.
2381 Frequency : SLES 9 only
2383 Description: Error of copying files with lustre special EAs as root
2384 Details : Client side always return success for setxattr call for lustre
2385 special xattr (currently only "trusted.lov").
2388 Frequency : rarely on clusters with both ia64+i386 clients
2390 Description: ia64+i686 clients doing shared IO on the same file may LBUG
2391 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
2392 clients are doing concurrent writes to the same file it is
2393 possible that the ia64 clients may LASSERT because the OST
2394 extent locks are not PAGE_SIZE aligned. Ensure that grown
2395 locks are always aligned on the request boundary.
2398 Frequency : specific use, occasional
2400 Description: Overwriting in use executable truncates on-disk binary image
2401 Details : If one node attempts to overwrite an executable in use by
2402 another node, we now correctly return ETXTBSY instead of
2403 truncating the file.
2405 Severity : enhancement
2407 Description: Async OSC create to avoid the blocking unnecessarily.
2408 Details : If a OST has no remain object, system will block on the creating
2409 when need to create a new object on this OST. Now, ways use
2410 pre-created objects when available, instead of blocking on an
2411 empty osc while others are not empty. If we must block, we block
2412 for the shortest possible period of time.
2417 Description: chmod on Lustre root is propagated to other clients
2418 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
2419 invalid by the follow_mount time.
2422 Frequency : liblustre clients only
2424 Description: Race in 'instant cancel' lock handling could lead to such locks
2425 never to be granted in case of SMP MDS
2426 Details : Do not destroy not yet granted but cbpending locks in
2430 Frequency : replay/resend of open
2432 Description: non null lock assetion failure in mds_intent_policy
2433 Details : Trying to replay/resend lockless open requests resulted in
2434 mds_open() returning 0 with no lock. Now it sets a flag if
2435 a lock is going to be returned.
2437 Severity : enhancement
2439 Description: Checksum enhancements
2440 Details : New checksum enhancements allow for resending RPCs that failed
2443 Severity : enhancement
2445 Description: Tunables on number of dirty pages in cacche
2446 Details : Allow to set limit on number of dirty pages cached.
2451 Description: client crash on unmount - lock still has references
2452 Details : In some error handling cases it was possible to leak a lock
2453 reference on a client while accessing a file. This was not
2454 harmful to the client during operation, but would cause the
2455 client to crash when the filesystem is unmounted.
2458 Frequency : specific case, rare
2460 Description: ETXTBSY on mds though file not in use by client
2461 Details : ETXTBSY is no longer incorrectly returned when attempting to
2462 chmod or chown a directory that the user previously tried to
2463 execute or a currently-executing binary.
2466 Frequency : extremely rare except on liblustre-based clients
2468 Description: Lustre space not freed when files are deleted
2469 Details : Clean up open-unlinked files after client eviction. Previously
2470 the unlink was skipped and the files remained as orphans.
2475 Description: OST failure "would be an LBUG" in waiting_locks_callback()
2476 Details : In some cases it was possible to send a blocking callback to
2477 a client doing a glimpse, even though that client didn't get
2478 a lock granted. When the glimpse lock is cancelled on the OST
2479 the freed lock is left on the waiting list and corrupted the list.
2482 Frequency : all core dumps
2484 Description: Broke core dumps to lustre
2485 Details : Negative dentry may be unhashed if parent does not have UPDATE
2486 lock, but some callers, e.g. do_coredump, expect dentry to be
2487 hashed after successful create, hash it in ll_create_it.
2489 ------------------------------------------------------------------------------
2491 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
2493 * Support for kernels:
2494 2.6.9-42.0.2.EL (RHEL 4)
2495 2.6.5-7.276 (SLES 9)
2496 2.4.21-40.EL (RHEL 3)
2497 2.6.12.6 vanilla (kernel.org)
2501 Frequency : always on RHEL 3
2503 Description: Number of open files grows over time
2504 Details : The number of open files grows over time, whether or not
2505 Lustre is started. This was due to a filp leak introduced
2506 by one of our kernel patches.
2508 ------------------------------------------------------------------------------
2510 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2512 * Support for kernels:
2513 2.6.9-42.EL (RHEL 4)
2514 2.6.5-7.276 (SLES 9)
2515 2.4.21-40.EL (RHEL 3)
2516 2.6.12.6 vanilla (kernel.org)
2521 Bugzilla : 5719, 9635, 9792, 9684
2522 Description: OST (or MDS) trips assertions in (re)connection under heavy load
2523 Details : If a server is under heavy load and cannot reply to new
2524 connection requests before the client resends the (re)connect,
2525 the connection handling code can behave badly if two service
2526 threads are concurrently handing separate (re)connections from
2527 the same client. Add better locking to the connection handling
2528 code, and ensure that only a single connection will be processed
2529 for a given client UUID, even if the lock is dropped.
2531 Severity : enhancement
2533 Description: add TCP zero-copy support to kernel
2534 Details : Add support to the kernel TCP stack to allow zero-copy bulk
2535 sends if the hardware supports scatter-gather and checksumming.
2536 This allows socklnd to do client-write and server-read more
2537 efficiently and reduce CPU utilization from skbuf copying.
2540 Frequency : only if NFS exporting from client
2542 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
2543 Details : If setting attributes on a file created under NFS that had
2544 never been opened it would be possible to oops the client
2545 if the file had no objects.
2548 Frequency : always for liblustre
2550 Description: liblustre client does MDS+OSTs setattr RPC for each write
2551 Details : When doing a write from a liblustre client, the client
2552 incorrectly issued an RPC to the MDS and each OST the file was
2553 striped over in order to update the timestamps. When writing
2554 with small chunks and many clients this could overwhelm the MDS
2555 with RPCs. In all cases it would slow down the write because
2556 these RPCs are unnecessary.
2558 Severity : enhancement
2560 Description: allow number of MDS service threads to be changed at module load
2561 Details : It is now possible to change the number of MDS service threads
2562 running. Adding "options mds mds_num_threads={N}" to the MDS's
2563 /etc/modprobe.conf will set the number of threads for the next
2564 time Lustre is restarted (assuming the "mds" module is also
2565 reloaded at that time). The default number of threads will
2566 stay the same, 32 for most systems.
2571 Description: OST crash if filesystem is unformatted or corrupt
2572 Details : If an OST is started on a device that has never been formatted
2573 or if the filesystem is corrupt and cannot even mount then the
2574 error handling cleanup routines would dereference a NULL pointer.
2579 Description: NULL pointer deref in llap_from_page.
2580 Details : get_cache_page_nowait can return a page with NULL (or otherwise
2581 incorrect) mapping if the page was truncated/reclaimed while it was
2582 searched for. Check for this condition and skip such pages when
2583 doing readahead. Introduce extra check to llap_from_page() to
2584 verify page->mapping->host is non-NULL (so page is not anonymous).
2587 Frequency : Sometimes when using sys_sendfile
2589 Description: "page not covered by a lock" warnings from ll_readpage
2590 Details : sendfile called ll_readpage without right page locks present.
2591 Now we introduced ll_file_sendfile that does necessary locking
2592 around call to generic_file_sendfile() much like we do in
2596 Frequency : with certain MDS communication failures at client mount time
2598 Description: NULL pointer deref after failed client mount
2599 Details : a client connection request may delayed by the network layer
2600 and not be sent until after the PTLRPC layer has timed out the
2601 request. If the client fails the mount immediately it will try
2602 to clean up before the network times out the request. Add a
2603 reference from the request import to the obd device and delay
2604 the cleanup until the network drops the request.
2607 Frequency : occasionally during client (re)connect
2609 Description: assertion failure during client (re)connect
2610 Details : processing a client connection request may be delayed by the
2611 client or server longer than the client connect timeout. This
2612 causes the client to resend the connection request. If the
2613 original connection request is replied in this interval, the
2614 client may trip an assertion failure in ptlrpc_connect_interpret()
2615 which thought it would be the only running connect process.
2618 Frequency : only with obd_echo servers and clients that are rebooted
2620 Description: kernel BUG accessing uninitialized data structure
2621 Details : When running an obd_echo server it did not start the ping_evictor
2622 thread, and when a client was evicted an uninitialized data
2623 structure was accessed. Start the ping_evictor in the RPC
2624 service startup instead of the OBD startup.
2626 Severity : enhancement
2627 Bugzilla : 10193 (patchless)
2628 Description: Remove dependency on various unexported kernel interfaces.
2629 Details : No longer need reparent_to_init, exit_mm, exit_files,
2630 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
2633 Frequency : rare (only users of deprecated and unsupported LDAP config)
2635 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
2636 Details : LDAP apparently contains 'lustreName' attributes instead of
2637 'name'. A simple remapping of the name is sufficient.
2640 Frequency : rare (only with non-default dump_on_timeout debug enabled)
2642 Description: waiting_locks_callback trips kernel BUG if client is evicted
2643 Details : Running with the dump_on_timeout debug flag turned on makes
2644 it possible that the waiting_locks_callback() can try to dump
2645 the Lustre kernel debug logs from an interrupt handler. Defer
2646 this log dumping to the expired_lock_main() thread.
2648 Severity : enhancement
2650 Description: Support NFS exporting on 2.6 kernels.
2651 Details : Implement non-rawops metadata methods for NFS server to use without
2652 changing NFS server code.
2655 Frequency : very rare (synthetic metadata workload only)
2657 Description: two racing renames might cause an MDS thread to deadlock
2658 Details : Running the "racer" program may cause one MDS thread to rename
2659 a file from being the source of a rename to being the target of
2660 a rename at exactly the same time that another thread is doing
2661 so, and the second thread has already enqueued these locks after
2662 doing a lookup of the target and is trying to relock them in
2663 order. Ensure that we don't try to re-lock the same resource.
2666 Frequency : only very large systems with liblustre clients
2668 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
2669 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
2670 all OSTs in parallel. This allows the eviction of stale liblustre
2671 clients to proceed much faster than if they were done in series,
2672 and also offers similar improvements for other set_info RPCs.
2677 Description: excessive CPU usage during initial read phase on client
2678 Details : During the initial read phase on a client, it would agressively
2679 retry readahead on the file, consuming too much CPU and impacting
2680 performance (since 1.4.5.8). Improve the readahead algorithm
2681 to avoid this, and also improve some other common cases (read
2682 of small files in particular, where "small" is files smaller than
2683 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
2688 Description: MDS crash when receiving packet with unknown intent.
2689 Details : Do not LBUG in unknown intent case, just return -EFAULT
2691 Severity : enhancement
2692 Bugzilla : 9293, 9385
2693 Description: MDS RPCs are serialised on client. This is unnecessary for some.
2694 Details : Do not serialize getattr (non-intent version) and statfs.
2697 Frequency : occasional, when OST network is overloaded/intermittent
2699 Description: client evicted by OST after bulk IO timeout
2700 Details : If a client sends a bulk IO request (read or write) the OST
2701 may evict the client if it is unresposive to its data GET/PUT
2702 request. This is incorrect if the network is overloaded (takes
2703 too long to transfer the RPC data) or dropped the OST GET/PUT
2704 request. There is no need to evict the client at all, since
2705 the pinger and/or lock callbacks will handle this, and the
2706 client can restart the bulk request.
2709 Frequency : Always when mmapping file with no objects
2711 Description: client crashes when mmapping file with no objects
2712 Details : Check that we actually have objects in a file before doing any
2713 operations on objects in ll_vm_open, ll_vm_close and
2719 Description: Request leak when working with deleted CWD
2720 Details : Introduce advanced request refcount tracking for requests
2721 referenced from lustre intent.
2723 Severity : Enhancement
2725 Description: Cache open file handles on client.
2726 Details : MDS now will return special lock along with openhandle, if
2727 requested and client is allowed to hold openhandle, even if unused,
2728 until such a lock is revoked. Helps NFS a lot, since NFS is opening
2729 closing files for every read/write openration.
2731 Severity : Enhancement
2733 Description: Cache open negative dentries on client when possible.
2734 Details : Guard negative dentries with UPDATE lock on parent dir, drop
2735 negative dentries on lock revocation.
2740 Description: Remounting a client read-only wasn't possible with a zconf mount
2741 Details : It wasn't possible to remount a client read-only with llmount.
2743 Severity : enhancement
2744 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
2745 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
2746 MPICH is included to add Lustre-specific ADIO interfaces.
2747 This is based closely on the UFS ADIO layer and only differs
2748 in file creation, in order to allow the OST striping to be set.
2749 This is user-contributed code and not supported by CFS.
2754 Description: extended inode attributes (immutable, append-only) work improperly
2755 when 2.4 and 2.6 kernels are used on client/server or vice versa
2756 Details : Introduce kernel-independent values for these flags.
2758 Severity : enhancement
2761 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
2762 Details : Many of the /proc/ tunables can only be tuned at a megabyte
2763 granularity. Now, Fractional MB granularity is be supported,
2764 this is very useful for low memory system.
2766 Severity : enhancement
2768 Description: Getattr by fid
2769 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
2770 locks, avoids extra getattr rpc requests to MDS, allows '/' to
2771 have locks and avoids getattr rpc requests for it on every stat.
2774 Frequency : Always, for filesystems larger than 2TB
2776 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2777 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
2778 filesystems larger than 2TB to be mounted with mballoc enabled.
2781 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
2783 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2784 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
2785 filesystems larger than 2TB is mounted then the kernel will
2786 silently corrupt the start of the filesystem. CONFIG_LBD is
2787 enabled for all CFS-supported kernels, but the possibility of
2788 this happening with a modified kernel config exists.
2790 Severity : enhancement
2792 Description: add client O_DIRECT support for 2.6 kernels
2793 Details : It is now possible to do O_DIRECT reads and writes to files
2794 in the Lustre client mountpoint on 2.6 kernel clients.
2796 Severity : enhancement
2798 Description: parallel glimpse, setattr, statfs, punch, destroy requests
2799 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
2800 parallel, not waiting for response from every OST before sending
2801 a rpc to the next OST.
2806 Description: setattr vs write race when updating file timestamps
2807 Details : Client processes that update a file timestamp into the past
2808 right after writing to the file (e.g. tar) it is possible that
2809 the updated file modification time can be reset to the current
2810 time due to a race between processing the setattr and write RPC.
2812 Severity : enhancement
2814 Description: Bring 'lfs find' closer in line with regular Linux find.
2815 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
2816 -print0 options and obtains all the needed info through the lustre
2819 Severity : enhancement
2821 Description: support up to 1024 configured devices on one node
2822 Details : change obd_dev array from statically allocated to dynamically
2823 allocated structs as they are first used to reduce memory usage
2828 Description: Flush dirty partially truncated pages during truncate
2829 Details : Immediatelly flush partially truncated pages in filter_setattr,
2830 this way we completely avoid having any pages in page cache on OST
2831 and can retire ugly workarounds during writes to flush such pages.
2836 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
2837 Details : protect inode from truncation within vfs_unlink() context
2838 just take a reference before calling vfs_unlink() and release it
2839 when parent's i_sem is free.
2844 Description: last_id value checked outside lock on OST caused LASSERT failure
2845 Details : If there were multiple MDS->OST object precreate requests in
2846 flight, it was possible that the OST's last object id was checked
2847 outside a lock and incorrectly tripped an assertion. Move checks
2848 inside locks, and discard old precreate requests.
2851 Frequency : always, if extents are used on OSTs
2853 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
2854 Details : OSTs using the extents format would not zero the high 16 bits of
2855 the index physical block number. This is not a problem for any
2856 OST filesystems smaller than 16TB, and no kernels support ext3
2857 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
2858 new/modified files) and can be fixed for existing filesystems
2859 with e2fsprogs-1.39-cfs1.
2864 Description: import connection selection may be incorrect if timer wraps
2865 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
2866 import connections to be ignored if the 32-bit jiffies counter
2867 wraps. Use a 64-bit jiffies counter.
2870 Frequency : very large clusters immediately after boot
2872 Description: LNET request buffers exhausted under heavy short-term load
2873 Details : If a large number of client requests are generated on a service
2874 that has previously never seen so many requests it is possible
2875 that the request buffer growth cannot keep up with the spike in
2876 demand. Instead of dropping incoming requests, they are held in
2877 the LND until the RPC service can accept more requests.
2880 Frequency : Sometimes during replay
2882 Description: Assertion failure in ll_local_open after replay.
2883 Details : If replay happened on an open request reply before we were able
2884 to set replay handler, reply will become not swabbed tripping the
2885 assertion in ll_local_open. Now we set the handler right after
2886 recognising of open request
2889 Frequency : very rare
2891 Description: kernel reports "badness in vsnprintf"
2892 Details : Reading from the "recovery_status" /proc file in small chunks
2893 may cause a negative length in lprocfs_obd_rd_recovery_status()
2894 call to vsnprintf() (which is otherwise harmless). Exit early
2895 if there is no more space in the output buffer.
2897 Severity : enhancement
2899 Description: clear OBD RPC statistics by writing to them
2900 Details : It is now possible to clear the OBD RPC statistics by writing
2901 to the "stats" file.
2906 Description: Client mtime is not the same on different clients after utimes
2907 Details : In some cases, the client was using the utimes() syscall on
2908 a file cached on another node. The clients now validate the
2909 ctime from the MDS + OSTs to determine which one is right.
2914 Description: Inability to activate failout mode
2915 Details : lconf script incorrectly assumed that in pythong string's numeric
2916 value is used in comparisons.
2919 Frequency : always with multiple stripes per file
2921 Description: Inefficient object allocation for mutli-stripe files
2922 Details : When selecting which OSTs to stripe files over, for files with
2923 a stripe count that divides evenly into the number of OSTs,
2924 the MDS is always picking the same starting OST for each file.
2925 Return the OST selection heuristic to the original design.
2930 Description: mount failures may take full timeout to return an error
2931 Details : Under some heavy load conditions it is possible that a
2932 failed mount can wait for the full obd_timeout interval,
2933 possibly several minutes, before reporting an error.
2934 Instead return an error as soon as the status is known.
2936 Frequency : quota enabled and large files being deleted
2938 Description: releasing more than 4GB of quota at once hangs OST
2939 Details : If a user deletes more than 4GB of files on a single OST it
2940 will cause the OST to spin in an infinite loop. Release
2941 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2946 Description: statfs data retrieved from /proc may be stale or zero
2947 Details : When reading per-device statfs data from /proc, in the
2948 {kbytes,files}_{total,free,avail} files, it may appear
2949 as zero or be out of date.
2952 Frequency : systems with MD RAID1 external journal devices
2954 Description: lconf's call to blkid is confused by RAID1 journal devices
2955 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2956 of returning all block devices that match the journal UUID.
2959 Frequency : always, for aggregate stripe size over 4GB
2961 Description: assertion fails when trying to use 4GB stripe size
2962 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
2963 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
2966 Frequency : always on ppc64
2968 Description: the first write on an ext3 filesystem with mballoc got stuck
2969 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2970 perform endianness conversion.
2972 ------------------------------------------------------------------------------
2974 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2976 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
2977 INTEROPERATE with older versions automatically. Please read the
2978 user documentation before upgrading any part of a live system.
2979 * WARNING: Lustre networking configuration changes are required with
2980 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
2983 * Support for kernels:
2984 2.6.9-22.0.2.EL (RHEL 4)
2985 2.6.5-7.244 (SLES 9)
2986 2.6.12.6 vanilla (kernel.org)
2989 Severity : enhancement
2990 Bugzilla : 7981/8208
2991 Description: Introduced Lustre Networking (LNET)
2992 Details : LNET is new networking infrastructure for Lustre, it includes
2993 a reorganized network configuration mode (see the user
2994 documentation for full details) as well as support for routing
2995 between different network fabrics. Lustre Networking Devices
2996 (LNDs) for the supported network fabrics have also been
2997 created for this new infrastructure.
2999 Severity : enhancement
3000 Description: Introduced Access control lists
3001 Details : clients can set ACLs on files and directories in order to have
3002 more fine-grained permissions than the standard Unix UGO+RWX.
3003 The MDS must be started with the "-o acl" mount option.
3005 Severity : enhancement
3006 Description: Introduced filesystem quotas
3007 Details : Administrators may now establish per-user quotas on the
3010 Severity : enhancement
3012 Description: Configuration change for the XT3
3013 The PTLLND is now used to run Lustre over Portals on the XT3
3014 The configure option(s) --with-cray-portals are no longer used.
3015 Rather --with-portals=<path-to-portals-includes> is used to
3016 enable building on the XT3. In addition to enable XT3 specific
3017 features the option --enable-cray-xt3 must be used.
3022 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
3023 Details : A race between allocating a new llcd and re-getting the llcd_lock
3024 allowed another thread to grab newly-allocated llcd.
3026 Severity : enhancement
3028 Description: 2.6 OST async journal commit and locking fix to improve performance
3029 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
3030 2.6 kernels are now async as they already were in 2.4 kernels so
3031 that they can commit concurrently with the network bulk transfer.
3032 For block-allocated files the filter allocation semaphore is held
3033 to avoid filesystem fragmentation during allocation. BKL lock
3034 removed for 2.6 xattr operations where it is no longer needed.
3039 Description: lconf incorrectly determined whether two IP networks could talk
3040 Details : In some more complicated routing and multiple-network
3041 configurations, lconf will avoid trying to make a network
3042 connection to a disjoint part of the IP space. It was doing the
3043 math incorrectly for one set of cases.
3048 Description: Fix for potential infinite loop processing records in an llog.
3049 Details : If an llog record is corrupted/zeroed, it is possible to loop
3050 forever in llog_process(). Validate the llog record length
3051 and skip the remainder of the block on error.
3054 Frequency : occasional (liblustre only)
3056 Description: liblustre could not open files whose last component is a symlink
3057 Details : sysio_path_walk() would incorrectly pass the open intent to
3058 intermediate path components.
3061 Frequency : rare (liblustre only with non-standard tuning)
3062 Bugzilla : 7201 (7350)
3063 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
3064 Details : llu_lookup_finish_locks() tries to set lock data on a lock
3065 after it has been released, only do this for referenced locks
3067 Severity : enhancement
3069 Description: specifying an (invalid) directory default stripe_size of -1
3070 would reset the directory default striping
3071 Details : stripe_size -1 was used internally to signal directory stripe
3072 removal, now use "all default" to signal dir stripe removal
3073 as a directory striping of "all default" is not useful
3076 Frequency : common for large clusters running liblustre clients
3078 Description: doing an ls when liblustre clients are running is slow
3079 Details : sending a glimpse AST to a liblustre client waits for every AST
3080 to time out, as liblustre clients will not respond. Since they
3081 cannot cache data we refresh the OST lock LVB from disk instead.
3083 Severity : enhancement
3085 Description: doing an ls at the same time as file IO can be slow
3086 Details : enqueue and other "small" requests can be blocked behind many
3087 large IO requests. Create a new OST IO portal for non-IO
3088 requests so they can be processed faster.
3091 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
3093 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
3094 created file with mode 000
3095 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
3096 allow an ftruncate() as a non-root user to a file with mode 000.
3097 Lustre now respects this flag to disable mode checks when
3098 truncating a file owned by the user
3101 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
3104 Description: Revoking locks from clients that went dead or catatonic might take
3106 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
3107 cancellation of such locks instant on servers without waiting for
3108 any reply from clients. Clients drops these locks when cancel
3109 notification from server is received without replying.
3112 Frequency : liblustre-only, when liblustre client dies or becomes busy
3114 Description: Doing ls on Linux clients can take a long time with active
3116 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
3117 granting such locks to it in the first place if possible. Locks
3118 are taken by proxy on the OST during the read or write and
3119 dropped immediately afterward. Add connect flags handling, do
3120 not grant locks to liblustre clients for glimpse ASTs.
3122 Severity : enhancement
3124 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
3125 Details : Existing read-ahead algorithm is tuned for the case of streamlined
3126 sequential reads and behaves badly with applications doing random
3127 reads. Improve it by reading ahead at least read region, and
3128 avoiding excessive large RPC for small reads.
3130 Severity : enhancement
3132 Description: Creating more than 1000 files for a single job may cause a load
3133 imbalance on the OSTs if there are also a large number of OSTs.
3134 Details : qos_prep_create() uses an OST index reseed value that is an
3135 even multiple of the number of available OSTs so that if the
3136 reseed happens in the middle of the object allocation it will
3137 still utilize the OSTs as uniformly as possible.
3142 Description: OST or MDS may oops in ping_evictor_main()
3143 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
3144 but doesn't restart at beginning of obd_exports_timed list
3147 Severity : enhancement
3149 Description: improve by-nid export eviction on the MDS and OST
3150 Details : allow multiple exports with the same NID to be evicted at one
3151 time without re-searching the exports list.
3154 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
3156 Description: MDS may oops in groups_free()
3157 Details : in rare race conditions a newly allocated group_info struct is
3158 freed again, and this can be NULL. The 2.4 compatibility code
3159 for groups_free() checked for a NULL pointer, but 2.6 did not.
3162 Frequency : common for liblustre clients doing little filesystem IO
3163 Bugzilla : 9352, 7313
3164 Description: server may evict liblustre clients accessing contended locks
3165 Details : if a client is granted a lock or receives a completion AST
3166 with a blocking AST already set it would not reply to the AST
3167 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
3170 Frequency : lfs setstripe, only systems with more than 160 OSTs
3172 Description: unable to set striping with a starting offset beyond OST 160
3173 Details : llapi_create_file() incorrectly limited the starting stripe
3174 index to the maximum single-file stripe count.
3177 Frequency : LDAP users only
3179 Description: lconf did not handle in-kernel recovery with LDAP properly
3180 Details : lconf/LustreDB get_refs() is searching the wrong namespace
3182 Severity : enhancement
3184 Description: bind OST threads to NUMA nodes to improve performance
3185 Details : all OST threads are uniformly bound to CPUs on a single NUMA
3186 node and do their allocations there to localize memory access
3188 Severity : enhancement
3190 Description: llmount can determine client NID directly from Myrinet (GM)
3191 Details : the client NID code from gmnalnid was moved directly into
3192 llmount, removing the need to use this or specifying the
3193 client NID explicitly when mounting GM clients with zeroconf
3196 Frequency : if client is started with down MDS
3198 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
3199 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
3200 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
3201 ptlrpc_import_delayed_req() because we didn't check if the
3202 request was interrupted, and we also didn't break out of the
3203 event loop if there was no timeout
3208 Description: data loss during non-page-aligned writes to a single file from
3209 both multiple nodes and multiple threads on one node at same time
3210 Details : updates to KMS and lsm weren't protected by common lock. Resulting
3211 inconsistency led to false short-reads, that were cached and later
3212 used by ->prepare_write() to fill in partially written page,
3213 leading to data loss.
3216 Frequency : always, if lconf --abort_recovery used
3218 Description: lconf --abort_recovery fails with 'Operation not supported'
3219 Details : lconf was attempting to abort recovery on the MDT device and not
3222 Severity : enhancement
3224 Description: remove cleanup logs
3225 Details : replace lconf-generated cleanup logs with lustre internal
3226 cleanup routines. Eliminates the need for client-cleanup and
3229 Severity : enhancement
3231 Description: add support for EAs (user and system) on lustre filesystems
3232 Details : it is now possible to store extended attributes in the Lustre
3233 client filesystem, and with the user_xattr mount option it
3234 is possible to allow users to store EAs on their files also
3236 Severity : enhancement
3238 Description: Add possibility (config option) to show minimal available OST free
3240 Details : When compiled with --enable-mindf configure option, statfs(2)
3241 (and so, df) will return least minimal free space available from
3242 all OSTs as amount of free space on FS, instead of summary of
3243 free spaces of all OSTs.
3245 Severity : enhancement
3247 Description: do not expand extent locks acquired on OST-side
3248 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
3249 by server: they are not cached anyway.
3252 Frequency : when mmap is used/binaries executed from Lustre
3254 Description: Unmmap pages before throwing them away from read cache.
3255 Details : llap_shrink cache now attempts to unmap pages before discarding
3256 them (if unmapping failed - do not discard). SLES9 kernel has
3257 extra checks that trigger if this unmapping is not done first.
3262 Description: lconf didn't resolve symlinks before checking to see whether a
3263 given mountpoint was already in use
3266 Frequency : when migrating failover services
3267 Bugzilla : 6395, 9514
3268 Description: When migrating a subset of services from a node (e.g. failback
3269 from a failover service node) the remaining services would
3270 time out and evict clients.
3271 Details : lconf --force (implied by --failover) sets the global obd_timeout
3272 to 5 seconds in order to quickly disconnect, but this caused
3273 other RPCs to time out too quickly. Do not change the global
3274 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
3276 Severity : enhancement
3277 Frequency : if MDS is started with down OST
3278 Bugzilla : 9439,5706
3279 Description: Allow startup/shutdown of an MDS without depending on the
3280 availability of the OSTs.
3281 Details : Asynchronously call mds_lov_synchronize during MDS startup.
3282 Add appropriate locking and lov-osc refcounts for safe
3283 cleaning. Add osc abort_inflight calls in case the
3284 synchronize never started.
3287 Frequency : occasional (Cray XT3 only)
3289 Description: root not authorized to access files in CRAY_PORTALS environment
3290 Details : The client process capabilities were not honoured on the MDS in
3291 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
3292 been accessed by an authorized user then root was able to access
3293 the file on the local client also. The root user capabilities
3294 are now allowed on the MDS, as this environment has secure UID.
3297 Frequency : occasional
3299 Description: ldiskfs "too long searching" message happens too often
3300 Details : A debugging message (otherwise harmless) prints too often on
3301 the OST console. This has been reduced to only happen when
3302 there are fragmentation problems on the filesystem.
3307 Description: Division by zero in statfs when all OSCs are inactive
3308 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
3309 lov_statfs divides by value returned by lov_get_stripecnt().
3313 Bugzilla : 9489, 3273
3314 Description: First write from each client to each OST was only 4kB in size,
3315 to initialize client writeback cache, which caused sub-optimal
3316 RPCs and poor layout on disk for the first writen file.
3317 Details : Clients now request an initial cache grant at (re)connect time
3318 and so that they can start streaming writes to the cache right
3319 away and always do full-sized RPCs if there is enough data.
3320 If the OST is rebooted the client also re-establishes its grant
3321 so that client cached writes will be honoured under the grant.
3326 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
3327 Details : Now I/O RPCs go to different portal number and (presumably) fast
3328 lock requests (and glimses) and other RPCs get their own service
3329 threads pool that should be able to service those RPCs
3332 Severity : enhancement
3334 Description: Ability to exchange lustre version between client and servers and
3335 issue warnings at client side if client is too old. Also for
3336 liblustre clients there is ability to refuse connection of too old
3338 Details : New 'version' field is added to connect data structure that is
3339 filled with version info. That info is later checked by server and
3343 Frequency : rare, liblustre only.
3344 Bugzilla : 9296, 9581
3345 Description: Two simultaneous writes from liblustre at offset within same page
3346 might proceed at the same time overwriting eachother with stale
3348 Details : I/O lock withing llu_file_prwv was released too early, before data
3349 actually was hitting the wire. Extended lock-holding time until
3350 server acknowledges receiving data.
3353 Frequency : extremely rare. Never observed in practice.
3355 Description: avoid generating lustre_handle cookie of 0.
3356 Details : class_handle_hash() generates handle cookies by incrementing
3357 global counter, and can hit 0 occasionaly (this is unlikely, but
3358 not impossible, because initial value of cookie counter is
3359 selected randonly). Value of 0 is used as a sentinel meaning
3360 "unassigned handle" --- avoid it. Also coalesce two critical
3361 sections in this function into one.
3363 Severity : enhancement
3365 Description: allow liblustre clients to delegate truncate locking to OST
3366 Details : To avoid overhead of locking, liblustre client instructs OST to
3367 take extent lock in ost_punch() on client's behalf. New connection
3368 flag is added to handle backward compatibility.
3370 Severity : enhancement
3371 Bugzilla : 4928, 7341, 9758
3372 Description: allow number of OST service threads to be specified
3373 Details : a module parameter allows the number of OST service threads
3374 to be specified via "options ost ost_num_threads={N}" in the
3375 OSS's /etc/modules.conf or /etc/modprobe.conf.
3379 Bugzilla : 6146, 9635, 9895
3380 Description: servers crash with bad pointer in target_handle_connect()
3381 Details : In rare cases when a client is reconnecting it was possible that
3382 the connection request was the last reference for that export.
3383 We would temporarily drop the export reference and get a new
3384 one, but this may have been the last reference and the export
3385 was just destroyed. Get new reference before dropping old one.
3387 Severity : enhancement
3388 Frequency : if client is started with failover MDS
3390 Description: Allow multiple MDS hostnames in the mount command
3391 Details : Try to read the configuration from all specified MDS
3392 hostnames during a client mount in case the "primary"
3395 Severity : enhancement
3397 Description: Stop sending data to evicted clients as soon as possible.
3398 Details : Check if the client we are about to send or are sending data to
3399 was evicted already. (Check is done every second of waiting,
3400 for which l_wait_event interface was extended to allow checking
3401 of exit condition at specified intervals).
3404 Frequency : rare, normally only when NFS exporting is done from client
3406 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
3408 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
3411 Frequency : always, if srand() is called before liblustre initialization
3413 Description: Liblustre uses system PRNG disturbing its usage by user application
3414 Details : Introduce internal to lustre fast and high-quality PRNG for
3415 lustre usage and make liblustre and some other places in generic
3416 lustre code to use it.
3418 Severity : enhancement
3419 Bugzilla : 9477, 9557, 9870
3420 Description: Verify that the MDS configuration logs are updated when xml is
3421 Details : Check if the .xml configuration logs are newer than the config
3422 logs stored on the MDS and report an error if this is the case.
3423 Request --write-conf, or allow starting with --old_conf.
3425 Severity : enhancement
3427 Description: Handle symlinks in the path when checking if Lustre is mounted.
3428 Details : Resolve intermediate symlinks when checking if a client has
3429 mounted a filesystem to avoid duplicate client mounts.
3434 Description: lconf can hit an error exception but still return success.
3435 Details : The lconf command catches the Command error exception at the top
3436 level script context and will exit with the associated exit
3437 status, but doesn't ensure that this exit status is non-zero.
3442 Description: failure of ptlrpc thread startup can cause oops
3443 Details : Starting a ptlrpc service thread can fail if there are a large
3444 number of threads or the server memory is very fragmented.
3445 Handle this without oopsing.
3448 Frequency : always, only if liblustre and non-default acceptor port was used
3450 Description: liblustre cannot connect to servers with non-default acceptor port
3451 Details : tcpnal_set_default_params() was not called and was therefore
3452 ignoring the environment varaible TCPNAL_PORT, as well as other
3453 TCPNAL_ environment variables
3458 Description: two objects could be created on the same OST for a single file
3459 Details : If an OST is down, in some cases it was possible to create two
3460 objects on a single OST for a single file. No problems other
3461 than potential performance impact and spurious error messages.
3465 Bugzilla : 5681, 9562
3466 Description: Client may oops in ll_unhash_aliases
3467 Details : Client dcache may become inconsistent in race condition.
3468 In some cases "getcwd" can fail if the current directory is
3474 Description: Inode refcounting problems in NFS export code
3475 Details : link_raw functions used to call d_instantiate without obtaining
3476 extra inode reference first.
3480 Bugzilla : 9942, 9903
3481 Description: Referencing freed requests leading to crash, memleaks with NFS.
3482 Details : We used to require that call to ll_revalidate_it was always
3483 followed by ll_lookup_it. Also with revalidate_special() it is
3484 possible to call ll_revalidate_it() twice for the same dentry
3485 even if first occurence returned success. This fix changes semantic
3486 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
3487 reference on a request referred from the intent.
3488 ll_intent_release() then releases such a request.
3491 Frequency : rare, normally benchmark loads only
3493 Description: unlinked inodes were kept in memory on the client
3494 Details : If a client is repeatedly creating and unlinking files it
3495 can accumulate a lot of stale inodes in the inode slab cache.
3496 If there is no other client load running this can cause the
3497 client node to run out of memory. Instead flush old inodes
3498 from client cache that have the same inode number as a new inode.
3501 Frequency : SLES9 2.6.5 kernel and long filenames only
3502 Bugzilla : 9969, 10379
3503 Description: utime reports stale NFS file handle
3504 Details : SLES9 uses out-of-dentry names in some cases, which confused
3505 the lustre dentry revalidation. Change it to always use the
3509 Frequency : rare, unless heavy write-truncate concurrency is continuous
3510 Bugzilla : 4180, 6984, 7171, 9963, 9331
3511 Description: OST becomes very slow and/or deadlocked during object unlink
3512 Details : filter_destroy() was holding onto the parent directory lock
3513 while truncating+unlinking objects. For very large objects this
3514 may block other threads for a long time and slow overall OST
3515 responsiveness. It may also be possible to get a lock ordering
3516 deadlock in this case, or run out of journal credits because of
3517 the combined truncate+unlink. Solution is to do object truncate
3518 first in one transaction without parent lock, and then do the
3519 final unlink in a new transaction with the parent lock. This
3520 reduces the lock hold time dramatically.
3523 Frequency : rare, 2.4 kernels only
3525 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
3526 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
3527 before unmounting their filesystem in order to avoid deadlock.
3528 The kernel_locked() function in 2.4 kernels only checks whether
3529 the kernel lock is held, not whether it is this process that is
3530 holding it as 2.6 kernels do.
3535 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
3536 Details : The client ptlrpc code may be trying to reconnect to a down
3537 server before a previous connection attempt has timed out.
3538 Increase the reconnect interval to be longer than the connection
3539 timeout interval to avoid sending duplicate connections to
3543 Frequency : echo_client brw_test command
3545 Description: fix echo_client to work with OST preallocated code
3546 Details : OST preallocation code (5137) didn't take echo_client IO path
3547 into account: echo_client calls filter methods outside of any
3548 OST thread and, hence, there is no per-thread preallocated
3549 pages and buffers to use. Solution: hijack pga pages for IO. As
3550 a byproduct, this avoids unnecessary data copying.
3554 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
3555 Description: Client can oops in mdc_commit_close() after open replay
3556 Details : It was possible for the MDS to return an open request with no
3557 transaction number in mds_finish_transno() if the client was
3558 evicted, but without actually returning an error. Clients
3559 would later try to replay that open and may trip an assertion
3560 Simplify the client close codepath, and always return an error
3561 from the MDS in case the open is not successful.
3564 Frequency : rare, 2.6 OSTs only
3566 Description: OST may deadlock under high load on fragmented files
3567 Details : If there was a heavy load and highly-fragmented OST filesystems
3568 it was possible to have all the OST threads deadlock waiting on
3569 allocation of biovecs, because the biovecs were not released
3570 until the entire RPC IO was completed. Instead, release biovecs
3571 as soon as they are complete to ensure forward IO progress.
3573 Severity : enhancement
3575 Description: Support for specifying external journal device at mount
3576 Details : If an OST or MDS device is formatted with an external journal
3577 device, this device major/minor is stored in the ext3 superblock
3578 and may not be valid for failover. Allow detecting and
3579 specifying the external journal at mount time.
3584 Description: Mounting an MDS with pending unlinked files may cause oops
3585 Details : target_finish_recovery() calls mds_postrecov() which returned
3586 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
3587 considers this an error and immediately begins cleaning up the
3588 lov, just after starting the mds_lov process
3590 Severity : enhancement
3592 Description: Implement 'lfs df' to report actual free space on per-OST basis
3593 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
3594 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
3595 usage of objects; '-h' to report in human readable format.
3597 ------------------------------------------------------------------------------
3599 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
3606 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
3607 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
3608 at startup during orphan unlinking. Instead, skip buddy update
3609 before setup, as it will be regenerated after recovery is complete.
3614 Description: If an OST is inactive, its locks might reference stale inodes.
3615 Details : lov_change_cbdata() must iterate over all namespaces, even if
3616 they are inactive to clear inode references from the lock.
3618 Severity : enhancement
3619 Frequency : occasional, if non-standard max_dirty_mb used
3621 Description: Client will block write RPCs if not enough grant
3622 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
3623 then the client will block writes while waiting for another RPC
3624 to complete instead of consuming its dirty limit. With change
3625 we get improved performance when max_dirty_mb is small.
3627 Severity : enhancement
3628 Bugzilla : 3389, 6253
3629 Description: Add support for supplementary groups on the MDS.
3630 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
3631 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
3632 lookups for user supplementary groups into a cache.
3636 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
3637 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
3638 for liblustre O_EXCL is enforced by MDS.
3642 Description: readdir never returns NULL in liblustre.
3643 Details : Corrected llu_iop_getdirentries logic, to return offset of next
3644 dentry in struct dirent.
3648 Frequency : liblustre only, depends on application IO pattern
3649 Description: liblustre clients evicted if not contacting servers
3650 Details : Don't put liblustre clients into the ping_evictor list, so
3651 they will not be evicted by the pinger ever.
3653 Severity : enhancement
3655 Description: Add ability to evict clients by NID from MDS.
3656 Details : By echoing "nid:$NID" string into
3657 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
3658 $NID would be instantly evicted from this MDS and from all active
3659 OSTs connected to it.
3663 Description: Do not query file size twice, somewhat slowing stat(2) calls.
3664 Details : lookup_it_finish() used to query file size from OSTs that was not
3669 Description: service threads change working directory to that of init
3670 Details : Starting lustre service threads may pin the working directory
3671 of the parent thread, making that filesystem busy. Threads
3672 now change to the working directory of init to avoid this.
3676 Frequency : during shutdown only
3677 Description: shutdown with a failed MDS or OST can cause unmount to hang
3678 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
3683 Frequency : 2.6 only
3684 Description: chmod/chown may include an extra supplementary group
3685 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
3686 supplementary group and if none is specified this is used.
3689 Bugzilla : 5479 (6816)
3691 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
3692 Details : If lookup is in progress on a file that is unlinked we might try
3693 to revalidate the inode and fail in revalidate after lookup is
3694 complete and ll_file_open() enqueues the open again but
3695 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
3698 Frequency : always, if lconf --abort_recovery used
3700 Description: lconf --abort_recovery fails with 'Operation not supported'
3701 Details : lconf was attempting to abort recovery on the MDT device and not
3704 ------------------------------------------------------------------------------
3706 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
3711 Frequency : rare (only unsupported configurations with a node running as an
3713 Bugzilla : 6514, 5137
3714 Description: Mounting a Lustre file system on a node running as an OST could
3716 Details : OSTs now preallocates memory needed to write out data at
3717 startup, instead of when needed, to avoid having to
3718 allocate memory in possibly low memory situations.
3719 Specifically, if the file system is mounted on on OST,
3720 memory pressure could force it to try to write out data,
3721 which it needed to allocate memory to do. Due to the low
3722 memory, it would be unable to do so and the node would
3723 become unresponsive.
3725 Severity : enhancement
3727 Description: Addition of lconf --service command line option
3728 Details : lconf now accepts a '--service <arg>' option, which is
3729 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
3731 Severity : enhancement
3733 Description: Failover mode is now the default for OSTs.
3734 Details : By default, OSTs will now run in failover mode. To return to
3735 the old behaviour, add '--failout' to the lmc line for OSTs.
3737 Severity : enhancement
3739 Description: Health checks are now provided for MDS and OSTs
3740 Details : Additional detailed health check information on MSD and OSTs
3741 is now provided through the procfs health_check value.
3744 Frequency : occasional, depends on IO load
3746 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
3747 after numerous create/delete cycles
3748 Details : The ext3 inode allocation policy would not allocate new inodes
3749 very well on the OSTs because there are no new directories
3750 being created. Instead we look for groups with free space if
3751 the parent directories are nearly full.
3756 Description: Network or server problems during mount may cause partially
3757 mounted clients instead of returning an error.
3758 Details : The config llog parsing code may overwrite the error return
3759 code during mount error handling, returning success instead
3765 Description: MDS can fail to allocate large reply buffers
3766 Details : After long uptimes the MDS can fail to allocate large reply
3767 buffers (e.g. zconf client mount config records) due to memory
3768 fragmentation or consumption by the buffer cache. Preallocate
3769 some large reply buffers so that these replies can be sent even
3770 under memory pressure.
3774 Frequency : rare (liblustre)
3775 Description: fsx running with liblustre complained that using truncate() to
3776 extend the file doesn't work. This patch corrects that issue.
3777 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
3778 ATTR_SIZE and lsm use in llu_setattr_raw.
3782 Frequency : rare, only 2.6 kernels
3783 Description: Unusual file access patterns on the MDS may result in inode
3784 data being lost in very rare circumstances.
3785 Details : Bad interaction between the ea-in-inode patch and the "no-read"
3786 code in the 2.6 kernel caused the inode and/or EA data not to
3787 be read from disk, causing single-file corruption.
3791 Frequency : rare, only 2.6 filesystems using extents
3792 Description: Heavy concurrent write and delete load may cause data corruption.
3793 Details : It was possible under high-load situations to have an extent
3794 metadata block in the block device cache from a just-unlinked
3795 file overwrite a newly-allocated data block. We now unmap any
3796 metadata buffers that alias just-allocated data blocks.
3800 Frequency : filesystems with default stripe_count larger than 77
3801 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
3802 Details : lconf specifies an inode size of 4096 bytes when the default
3803 stripe_count is larger than 77. This conflicts with the default
3804 inode density of 1 per 4096 bytes. Allocate smaller inodes in
3805 this case to avoid pinning too much memory for large EAs.
3807 ------------------------------------------------------------------------------
3809 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
3814 Frequency : rare (extremely heavy IO load with hundreds of clients)
3816 Description: Client is evicted, gets IO error writing to file
3817 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
3818 caused clients to be evicted for AST timeouts. The fixes in
3819 bug 5192 mean we no longer need to have such short AST timeouts
3820 so ldlm_timeout has been increased.
3823 Frequency : occasional during --force or --failover shutdown under load
3824 Bugzilla : 5949, 4834
3825 Description: Server oops/LBUG if stopped with --force or --failover under load
3826 Details : a collection of import/export refcount and cleanup ordering
3827 issues fixed for safer force cleanup
3830 Frequency : only filesystems larger than 120 OSTs
3831 Bugzilla : 5990, 6223
3832 Description: lfs getstripe would oops on a very large filesystem
3833 Details : lov_getconfig used kfree on vmalloc'd memory
3836 Frequency : only filesystems exporting via NFS to Solaris 10 clients
3837 Bugzilla : 6242, 6243
3838 Description: reading from files that had been truncated to a non-zero size
3839 but never opened returned no data
3840 Details : ll_file_read() reads zeros from no-object files to EOF
3845 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
3846 Details : There's little harm in aborting MDS/OSS recovery and letting it
3847 try again, so I removed the LASSERT and return an error instead.
3849 Severity : enhancement
3851 Description: New debugging infrastructure for tracking down data corruption
3852 Details : The I/O checksum code was replaced to: (a) control it at runtime,
3853 (b) cover more of the client-side code path, and (c) try to narrow
3854 down where problems occurred
3858 Bugzilla : 3819, 4364, 4397, 6313
3859 Description: Racing close and eviction MDS could cause assertion in mds_close
3860 Details : It was possible to get multiple mfd references during close and
3861 client eviction, leading to one thread referencing a freed mfd.
3863 Severity: : enhancement
3864 Bugzilla : 3262, 6359
3865 Description: Attempts to reconnect to servers are now more aggressive.
3866 Details : This builds on the enhanced upcall-less recovery that was added
3867 in 1.4.2. When trying to reconnect to servers, clients will
3868 now try each server in the failover group every 10 seconds. By
3869 default, clients would previously try one server every 25 seconds.
3874 Description: After recovery, certain operations trigger a failed
3875 assertion on a client.
3876 Details : Failing over an mds, using lconf -d --failover, while a
3877 client was doing a readdir() call would cause the client to
3878 LBUG after recovery completed and the readdir() was resent.
3880 Severity : enhancement
3882 Description: Default groups are now added by lconf
3883 Details : You can now run lconf --group <servicename> without having to
3884 manually add groups with lmc.
3887 Frequency : occasional
3889 Description: Nodes with an elan id of 0 trigger a failed assertion
3892 Frequency : always when accessing e.g. tty/console device nodes
3894 Description: tty and some other devices nodes cannot be used on lustre
3895 Details : file's private_data field is used by device data and lustre
3896 values in there got lost. New field was added to struct file to
3897 store fs-specific private data.
3900 Frequency : when exporting Lustre via NFS
3902 Description: NFSD failed occasionally when looking up a path component
3903 Details : NFSD is looking up ".." which was broken in ext3 directories
3904 that had grown large enough to become hashed.
3907 Frequency : Clusters with multiple interfaces not on the same subnet
3909 Description: Nodes will repeatedly try to reconnect to an interface which it
3910 cannot reach and report an error to the log.
3911 Details : Extra peer list entries will be created by lconf with some peers
3912 unreachable. lconf now validates the peer before adding it.
3915 Frequency : Only if a default stripe is set on the filesystem root.
3917 Description: Setting a default stripe on the filesystem root prevented the
3918 filesystem from being remounted.
3919 Details : The client was sending extra request flags in the root getattr
3920 request and did not allocate a reply buffer for the dir EA.
3923 Frequency : occasional, higher if lots of files are accessed by one client
3924 Bugzilla : 6159, 6097
3925 Description: Client trips assertion regarding lsm mismatch/magic
3926 Details : While revalidating inodes the VFS looks up inodes with ifind()
3927 and in rare cases can find an inode that is being freed.
3928 The ll_test_inode() code will free the lsm during ifind()
3929 when it finds an existing inode and then the VFS later attaches
3930 this free lsm to a new inode.
3934 Bugzilla : 6422, 7030
3935 Description: MDS deadlock between mkdir and client eviction
3936 Details : Creating a new file via mkdir or mknod (starting a transaction
3937 and getting the ns lock) can deadlock with client eviction
3938 (gets ns lock and trying to finish a synchronous transaction).
3941 Frequency : occasional
3942 Description: While starting a server, the fsfilt_ext3 module could not be
3944 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
3945 kernels. Previously, lconf would still use the ext3 name
3946 when trying to load modules. Now, it will correctly use
3947 ext3 on 2.4 and ldiskfs on 2.6.
3949 Severity : enhancement
3950 Description: The default stripe count has been changed to 1
3951 Details : The interpretation of the default stripe count (0, to lfs
3952 or lmc) has been changed to mean striping across a single
3953 OST, rather than all available. For general usage we have
3954 found a stripe count of 1 or 2 works best.
3956 Severity : enhancement
3957 Description: Add support for compiling against Cray portals.
3958 Details : Conditional compiling for some areas that are different
3962 Frequency : occasional
3963 Bugzilla : 6409, 6834
3964 Description: Creating files with an explicit stripe count may lead to
3965 a failed assertion on the MDS
3966 Details : If some OSTs are full or unavailable, creating files may
3967 trigger a failed assertion on the MDS. Now, Lustre will
3968 try to use other servers or return an error to the
3972 Frequency : occasional
3974 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
3975 caused lock timeout during readahead (since 1.4.2).
3976 Details : Processes doing readahead might match a lock that hasn't been
3977 granted yet if there are overlapping and conflicting lock
3978 requests. The readahead process waits on ungranted lock
3979 (original lock is CBPENDING), while OST waits for that process
3980 to cancel CBPENDING read lock and eventually evicts client.
3982 Severity : enhancement
3984 Description: Initial enabling of flock support for clients
3985 Details : Implements fcntl advisory locking and file status functions.
3986 This feature is provided as an optional mount flag (default
3987 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
3988 locking are implemented yet, and those that are are not guaranteed
3989 to be completely correct in production environments.
3990 mount -t lustre -o [flock|noflock] ...
3993 Frequency : occasional
3995 Description: OSTs running 2.4 kernels but with extents enabled might trip an
3996 assertion in the ext3 JBD (journaling) layer.
3997 Details : The b_committed_data struct is protected by the big kernel lock
3998 in 2.4 kernels, serializing journal_commit_transaction() and
3999 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
4000 access to this struct. In 2.6 kernels there is finer grained
4001 locking to improve SMP performance of the JBD layer.
4005 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
4010 Description: merge and cleanup kernel patches.
4013 -----------------------------------------------------------------------------
4015 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
4017 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
4018 versions of Lustre. Please update all servers and clients to
4019 version 1.4.2 or later at the same time. You must also run
4020 "lconf --write-conf {config}.xml" on the MDS while it is stopped
4021 to update the configuration logs.
4023 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
4024 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
4025 - swab llog records of type '0' so we get proper header size/idx (5861)
4026 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
4027 - fix rename of one directory over another leaking an inode (5953)
4028 - avoid SetPageDirty on 2.6 (5981)
4029 - don't re-add just-being-destroyed locks to the waiting list (5653)
4030 - when creating new directories, inherit the parent's custom
4031 striping settings if present parent (3048)
4032 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
4033 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
4034 - don't hold client locks on temporary worklist from l_lru (5666)
4035 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
4036 - automatically evict dead clients (5921)
4037 - Update file size properly in create+truncate+fstat case (6196)
4038 - Do not unhash mountpoint dentries, do not allow removal of
4040 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
4041 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
4042 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
4043 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
4044 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
4045 - Correctly handle reads of files with no objects (6243)
4046 - lctl recover will also mark a device active if deactivate used (5933)
4048 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
4049 - allow --write-conf on an MDS with different nettype than client (5619)
4050 - don't write config llogs to MDS for mounts not from that MDS (5617)
4051 - lconf should create multiple TCP connections from a client (5201)
4052 - init scripts are now turned off by default; run chkconfig --on
4053 lustre and chkconfig --on lustrefs to use them
4054 - upcalls are no longer needed for clients to recover to failover
4056 - add --abort-recovery option to lconf to abort recovery on device
4058 - add support for an arbitrary number of OSTs (3026)
4059 - Quota support protocol changes.
4060 - forward compatibility changes to wire structs (6007)
4061 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
4062 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
4063 - improved "lustre status" script
4064 - initialize blocksize for non-regular files (6062)
4065 - added --disable-server and --disable-client configure options (5782)
4066 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
4067 - Vanilla 2.4.29 support
4068 - increase maximum number of obd devices to 520 (6242)
4069 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
4070 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
4071 SLES 9 2.6.5 kernels
4072 - limit stripes per file to 160 (the maximum EA size) (6093)
4074 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
4077 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
4078 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
4079 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
4080 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
4081 - small ext3 extents cleanups and fixes (5733)
4082 - improved mballoc code, several small races and bugs fixed (5733, 5638)
4083 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
4084 - avoid needless client->OST connect, fix handle mismatch (5317)
4085 - fix DLM error path that led to out-of-sync client, long delays (5779)
4086 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
4087 - fix several locking issues related to i_size (5492,5624,5654,5672)
4088 - don't move pending lock onto export if it is already evicted (5683)
4089 - fix kernel oops when creating .foo in unlinked directory (5548)
4090 - fix deadlock in obdfilter statistics vs. object create (5811)
4091 - use time_{before,after} to avoid timer jiffies wrap (5882)
4092 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
4093 - Do not leak request if server was not able to process it (5154)
4094 - If mds_open unable to find parent dir, make that negative lookup(5154)
4095 - don't create new directories with extent-mapping (5909, 5936)
4097 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
4098 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
4099 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
4100 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
4101 - support for RHEL4 kernel on i686 (5773)
4102 - provide error messages when incompatible logs are encountered (5898)
4104 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
4105 * version 1.4.0.10 (1.4.1 release candidate 1)
4107 - don't keep a lock reference when lock is not granted (4238)
4108 - unsafe list practices (rarely) led to infinite eviction loop (4908)
4109 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
4110 - drop import inflight refcount on signal_completed_replay error (5255)
4111 - unlock page after async write error during send (3677)
4112 - handle missing objects in filter_preprw_read properly (5265)
4113 - no transno return for symlink open, don't save no-trasno open (3440)
4114 - don't try to complete elan receive that already failed (4012)
4115 - free RPC server reply state on error (5406)
4116 - clean up thread from ptlrpc_start_thread() on error (5160)
4117 - readahead could read extra page into cache that wasn't ejected (5388)
4118 - prevent races in class_attach/setup/cleanup/detach (5260)
4119 - don't dereference de->d_inode after l_dput of de (5458)
4120 - use "int" for stripe value returned from lock_to_stripe (5544)
4121 - mballoc allocation and error-checking fixes in 2.6 (5504)
4122 - block device patches to fix I/O request sizes in 2.6 (5482)
4123 - look up hostnames for IB nals (5602)
4124 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
4125 - don't start multiple acceptors for the same port (5277)
4126 - fix incorrect LASSERT in mds_getattr_name (5635)
4127 - export a proc file for general "ping" checking (5628)
4128 - fix "lfs check" to not block when the MDS is down (5628)
4130 - service request history (4965)
4131 - put {ll,lov,osc}_async_page structs in a single slab (4699)
4132 - create an "evict_client" /proc entry on OSTs, like the MDS has
4133 - fix mount usage message, return errors per mount(8) (5168)
4134 - change grep [] to grep "[]" in tests so they work in more UMLs
4135 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
4136 - remove ancient LOV_MAGIC_V0 EA support (5047)
4137 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
4138 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
4139 - allow readahead allocations to fail when low on memory (5383)
4140 - mmap locking landed again, after considerable improvement (2828)
4141 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
4143 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
4146 - send OST transaction number in read/write reply to free req (4966)
4147 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
4148 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
4149 - the watchdog thread now runs as interruptible (5246)
4150 - flock/lockf fixes (but it's still disabled, pending 5135)
4151 - don't use EXT3 constants in llite code (5094)
4152 - memory shortage at startup could cause assertion (5176)
4154 - reorganization of lov code
4155 - single portals codebase
4157 - add extents/mballoc support (5025)
4158 - direct I/O reads in the obdfilter (4048)
4159 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
4161 tbd Cluster File Systems, Inc. <info@clusterfs.com>
4164 - send OST transaction number in read/write reply to free req (4966)
4165 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
4166 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
4167 - the watchdog thread now runs as interruptible (5246)
4168 - handle missing objects in filter_preprw_read properly (5265)
4169 - unsafe list practices (rarely) led to infinite eviction loop (4908)
4170 - drop import inflight refcount on signal_completed_replay error (5255)
4171 - unlock page after async write error during send (3677)
4172 - return original error code on reconstructed replies (3761)
4173 - no transno return for symlink open, don't save no-trasno open (3440)
4175 - add pid to ldlm debugging output (4922)
4176 - bump the watchdog timeouts -- we can't handle 30sec yet
4177 - extra debugging for orphan dentry/inode bug (5259)
4179 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
4182 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
4183 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
4184 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
4185 - fix NULL dereference in /proc/sys/portals/routes (4827)
4186 - allow failed mdc_close() operations to be interrupted (4561)
4187 - stop precreate on OST before MDS would time out on it (4778)
4188 - don't send partial-page writes before EOF from client (4410)
4189 - discard client grant for sub-page writes on large-page clients (4520)
4190 - don't free dentries not owned by NFS code, check generation (4806)
4191 - fix lsm leak if mds_create_objects() fails (4801)
4192 - limit debug_daemon file size, always print CERROR messages (4789)
4193 - use transno after validating reply (3892)
4194 - process timed out requests if import state changes (3754)
4195 - update mtime on OST during writes, return in glimpse (4829)
4196 - add mkfsoptions to LDAP (4679)
4197 - use ->max_readahead method instead of zapping global ra (5039)
4198 - don't interrupt __l_wait_event() during strace
4200 - add software watchdogs to catch hung threads quickly (4941)
4201 - make lustrefs init script start after nfs is mounted
4202 - fix CWARN/ERROR duplication (4930)
4203 - return async write errors to application if possible (2248)
4204 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
4205 - print NAL number in %x format (4645)
4206 - update barely-supported suse-2.4.21-171 series (4842)
4207 - support for sles 9 %post scripts
4208 - support for building 2.6 kernel-source packages
4209 - support for sles km_* packages
4211 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
4214 - ignore -ENOENT errors in osc_destroy (3639)
4215 - notify osc create thread that OSC is being cleaned up (4600)
4216 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
4217 - reconstruct ost_handle() like mds_handle() (4657)
4218 - create a new thread to do import eviction to avoid deadlock (3969)
4219 - let lconf resolve symlinked-to devices (4629)
4220 - don't unlink "objects" from directory with default EA (4554)
4221 - hold socknal file ref over connect in case target is down (4394)
4222 - allow more than 32000 subdirectories in a single directory (3244)
4223 - fix blocks count for O_DIRECT writes (3751)
4224 - OST returns ENOSPC from object create when no space left (4539)
4225 - don't send truncate RPC if file size isn't changing (4410)
4226 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
4227 - bind to privileged port in socknal and tcpnal (3689)
4229 - rate limit CERROR/CWARN console message to avoid overload (4519)
4230 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
4231 - basic mmap support (3918)
4232 - kernel patch series update from b1_4 (4711)
4234 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
4237 - avoid crash during MDS cleanup with OST shut down (2775)
4238 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
4239 - don't use bad inodes on the MDS (3744)
4240 - dynamic object preallocation to improve recovery speed (4236)
4241 - don't hold spinlock over lock dumping or change debug flags (4401)
4242 - don't zero obd_dev when it is force cleaned (3651)
4243 - print grants to console if they go negative (4431)
4244 - "lctl deactivate" will stop automatic recovery attempts (3406)
4245 - look for existing locks in ldlm_handle_enqueue() (3764)
4246 - don't resolve lock handle twice in recovery avoiding race (4401)
4247 - revalidate should check working dir is a directory (4134)
4249 - don't always mark "slow" obdfilter messages as errors (4418)
4251 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
4254 - don't close LustreDB during write_conf until it is done (3860)
4255 - fix typo in lconf for_each_profile (3821)
4256 - allow dumping logs from multiple threads at one time (3820)
4257 - don't allow multiple threads in OSC recovery (3812)
4258 - fix debug_size parameters (3864)
4259 - fix mds_postrecov to initialize import for llog ctxt (3121)
4260 - replace config semaphore with spinlock (3306)
4261 - be sure to send a reply for a CANCEL rpc with bad export (3863)
4262 - don't allow enqueue to complete on a destroyed export (3822)
4263 - down write_lock before checking llog header bitmap (3825)
4264 - recover from lock replay timeout (3764)
4265 - up llog sem before sending rpc (3652)
4266 - reduce ns lock hold times when setting kms (3267)
4267 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
4268 - fix NULL deref and obd_dev leak on setup error (3312)
4269 - replace some LBUG about llog ops with error handling (3841)
4270 - don't match INVALID dentries from d_lookup and spin (3784)
4271 - hold dcache_lock while marking dentries INVALID and hashing (4255)
4272 - fix invalid assertion in ptlrpc_set_wait (3880)
4274 - add libwrap support for the TCP acceptor (3996)
4275 - add /proc/sys/portals/routes for non-root route listing (3994)
4276 - allow setting MDS UUID in .xml (2580)
4277 - print the stack of a process that LBUGs (4228)
4279 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
4282 - don't cleanup request in ll_file_open() on failed MDS open (3430)
4283 - make sure to unset replay flag from failed open requests (3440)
4284 - if default stripe count is 0, use OST count for inode size (3636)
4285 - update parent mtime/ctime on client for create/unlink (2611)
4286 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
4287 - free recovery state on server during a forced cleanup (3571)
4288 - unregister_reply for resent reqs (3063)
4289 - loop back devices mounting and status check on 2.6 (3563)
4290 - fix resource-creation race that can provoke i_size == 0 (3513)
4291 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
4292 - more debugging for page-accounting assertion (3746)
4293 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
4294 - avoid deadlock after precreation failure (3758)
4295 - fix race and lock order deadlock in orphan handling (3450, 3750)
4296 - add validity checks when grabbing inodes from l_ast_data (3599)
4298 - add /proc/.../recovery_status to obdfilter (3428)
4299 - lightweight CDEBUG infrastructure, debug daemon (3668)
4300 - change default OSC RPC parameters to be better on small clusters
4301 - turn off OST read cache for files smaller than 32MB
4302 - install man pages and include them in rpms (3100)
4303 - add new init script for (un)mounting lustre filesystems (2593)
4304 - run chkconfig in %post for init scripts (3701)
4305 - drop scimac NAL (unmaintained)
4307 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
4310 - clean kiobufs before and after use (3485)
4311 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
4312 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
4313 - change init script to fail more clearly if not run as root (1528)
4314 - allow clients to reconnect during replay (1742)
4315 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
4316 - don't do DNS lookups on NIDs too small for IP addresses (3442)
4317 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
4318 - fix cond_resched (3554)
4319 - only evict unfinished clients after recovery (3515)
4320 - allow bulk resend, prevent data loss (3570)
4321 - dynamic ptlrpc request buffer allocation (2102)
4322 - don't allow unlinking open directory if it isn't empty (2904)
4323 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
4324 - remove extraneous obd dereference causing LASSERT failure (3334)
4325 - don't use get_cycles() when creating temp. files on the mds (3156)
4326 - hold i_sem when setting i_size in ll_extent_lock() (3564)
4327 - handle EEXIST for set-stripe, set proper directory name (3336)
4329 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
4330 - fix ksocknal_fmb_callback() error messages (2918)
4332 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
4335 - don't copy lvb into (possibly NULL) reply on error (2983)
4336 - don't deref dentry after dput, don't free lvb on error (2922)
4337 - use the kms to determine writeback rpc length (2947)
4338 - increment oti_logcookies when osc is inactive (2948)
4339 - update client's i_blocks count via lvb messages (2543)
4340 - handle intent open/close of special files properly (1557)
4341 - mount MDS with errors=remount-ro, like obdfilter (2009)
4342 - initialize lock handle to avoid ASSERT on error cleanup (3057)
4343 - don't use cancelling-locks' kms values (2947)
4344 - use highest lock extent for kms, not last one (2925)
4345 - don't dereference ERR_PTR() dentry in error handling path (3107)
4346 - fix thread race in portals_debug_dumplog() (3122)
4347 - create lprocfs device entries at setup instead of at attach (1519)
4348 - common AST error handler, don't evict client on completion race (3145)
4349 - zero nameidata in detach_mnt in 2.6 (3118)
4350 - verify d_inode after revalidate_special is valid in 2.6 (3116)
4351 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
4352 - initialize RPC timeout timer earlier for 2.6 (3219)
4353 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
4354 - print nal/nid for unknown nid (3258)
4355 - additional checks for oscc recovery before doing precreate (3284)
4356 - fix ll_extent_lock() error return code for 64-bit systems (3043)
4357 - don't crash in mdc_close for bad permissions on open (3285)
4358 - zero i_rdev for non-device files (3147)
4359 - clear page->private before handing to FS, better assertion (3119)
4360 - tune the read pipeline (3236)
4361 - fix incorrect decref of invalidated dentry (2350)
4362 - provide read-ahead stats and refine rpc in flight stats (3328)
4363 - don't hold journal transaction open across create RPC (3313)
4364 - update atime on MDS at close time (3265)
4365 - close LDAP connection when recovering to avoid server load (3315)
4366 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
4367 - don't leak open file on MDS after open resend (3325)
4368 - serialize filter_precreate and filter_destroy_precreated (3329)
4369 - loop device shouldn't call sync_dev() for nul device (3092)
4370 - clear page cache after eviction (2766)
4371 - resynchronize MDS->OST in background (2824)
4372 - refuse to mount the same filesystem twice on same mountpoint (3394)
4373 - allow llmount to create routes for mounting behind routers (3320)
4374 - push lock cancellation to blocking thread for glimpse ASTs (3409)
4375 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
4376 - fix rare problem with rename on htree directories (3417)
4378 - allow default OST striping configuration per directory (1414)
4379 - fix compilation for qswnal for 2.6 kernels (3125)
4380 - increase maximum number of MDS request buffers for large systems
4381 - change liblustreapi to be useful for external progs like lfsck (3098)
4382 - increase local configuration timeout for slow disks (3353)
4383 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
4385 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
4388 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
4389 - don't overwrite extent policy data in reply if lock was blocked (2901)
4390 - drop filter export grants atomically with removal from device (2663)
4391 - del obd_self_export from work_list in class_disconnect_exports (2908)
4392 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
4393 - swab reply message in mdc_close, other PPC fixes (2464)
4394 - fix destroying of named logs (2325)
4395 - overwrite old logs when running lconf --write_conf (2264)
4396 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
4397 - fix race in target_handle_connect (2898)
4398 - mds_reint_create() should take same inode create lock (2926)
4399 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
4400 - don't close files for self_export to avoid uninitialized obd (2936)
4401 - allow MDS with the same name as client node (2939)
4402 - hold dentry reference for closed log files for unlink (2325)
4403 - reserve space for all logs during transactions (2059)
4404 - don't evict page beyond end of stripe extent (2925)
4405 - don't oops on a deleted current working directory (2399)
4406 - handle hard links to targets without a parent properly (2517)
4407 - don't dereference NULL lock when racing during eviction (2867)
4408 - don't grow lock extents when lots of conflicting locks (2919)
4410 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
4413 - account for cache space usage on clients to avoid data loss (974)
4414 - lfsck support in lustre kernel code (2349)
4415 - reduce journal credits needed for BRW writes (2370)
4416 - orphan handling to avoid losing space on client/server crashes
4417 - ptlrpcd can be blocked, stopping ALL progress (2477)
4418 - use lock value blocks to assist in proper KMS, faster stat (1021)
4419 - takes i_sem instead of DLM locks internally on obdfilter (2720)
4420 - recovery for initial connections (2355)
4421 - fixes for mds_cleanup_orphans (1934)
4422 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
4423 - block all file creations until orphan recovery completes (1901)
4424 - client remove rq_connection from request struct (2423)
4425 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
4426 - recovery timer race (2670)
4427 - mdc_close recovey bug (2532)
4428 - ptlrpc cleanup bug (2710)
4429 - mds timeout on local locks (2588)
4430 - namespace lock held during RPCs (2431)
4431 - handle interrupted sync write properly (2503)
4432 - don't try to handle a message that hasn't been replied to (2699)
4433 - client assert failure during cleanup after abort recovery (2701)
4434 - leak mdc device after failed mount (2712)
4435 - ptlrpc_check_set allows timedout requests to complete (2714)
4436 - wait for inflight reqs when ptlrpcd finishes (2710)
4437 - make sure unregistered services are removed from the srv_list
4438 - reset bulk XID's when resending them (caught by 1138 test)
4439 - unregister_bulk after timeout
4440 - fix lconf error (2694)
4441 - handle write after unfinished setstripe, stripe-only getstripe (2388)
4442 - readahead locks pages, leaves pending causing memory pressure (2673)
4443 - increase OST request buffers to 4096 on large machines (2729)
4444 - fix up permission of existing directories in simple_mkdir (2661)
4445 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
4446 - don't assign transno to errored transactions (2742)
4447 - don't delete objects on OST if given a bogus objid from MDS (2751)
4448 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
4449 - if rq_no_resend, then timeout request after recovery (2432)
4450 - fix MDS llog_logid record size, 64-bit array alignment (2733)
4451 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
4452 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
4453 - MDS recovery shouldn't delete objects that it has given out (2730)
4454 - if enqueue arrives after completion, don't clobber LVB (2819)
4455 - don't unlock pages twice when trigger_group_io returns error (2814)
4456 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
4457 - don't write pages to disk if there was an error (1450)
4458 - don't ping imports that have recovery disabled (2676)
4459 - take buffered bytes into account when balancing socknal conn (2817)
4460 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
4461 - reconnect unlink llog connection after MDS reconnects to OST (2816)
4462 - remove little-endian swabbing of llog records (1987)
4463 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
4464 - retry reposting request buffers if they fail (1191)
4465 - grow extent at grant time to avoid granting a revoked lock (2809)
4466 - lock revoke doesn't evict page if covered by a second lock (2765)
4467 - disable VM readahead to avoid reading outside lock extents (2805)
4469 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
4470 - updated kernel patches for hp-2.4.20 kernel (2681)
4472 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
4475 - fix truncated write corruption (2366)
4476 - fix for failed assertion in iopen_connect_dentry (1792,2517)
4478 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
4479 - ensure len(uuid) < 37 in lmc (1171)
4480 - fix ia64 OOPS in llog_test (2255)
4481 - zero end of page at obdfilter for partial page writes (2648)
4482 - don't leave stale dentries around after renames (bug 2428)
4483 - fix timeouts when evicting a client with a single lock held (2642)
4484 - set deadline for the initial HELLO message to drain (2634)
4485 - print out dotted-quad IP addresses in the socknal (2302)
4487 - additional debugging for MDS client eviction problem (2443)
4488 - fix mkfsoptions support for osts (2603, 2604)
4490 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
4493 - add series for the vanilla 2.6.0 kernel
4494 - add series for the vanilla 2.4.24 kernel
4495 - add series for a cray x86/64 UL kernel drop
4496 - fix xattr patches for the vanilla 2.4.19 series
4498 - generate true UUIDs in lmc (1171)
4499 - have portals stack dumping break in UML (2466)
4500 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
4501 - allocate with _NOFS in ldlm to avoid deadlock (1933)
4502 - wake callback waiting threads on client eviction (2460)
4503 - Add --ptldebug and --subsystem to lmc (1719)
4504 - update assertion to allow safe interrupt allocation
4505 - set rq_no_resend for cancel requests (2432)
4506 - recalculate ptlrpcd timeout after resend (2494)
4507 - call vfs_rmdir when removing pending directories (2368)
4508 - fix renaming a file to itself (2429)
4509 - lmc creates a default one-stripe lov (2454)
4510 - expand procfs space to handle large clusters (2326)
4511 - increase UML stack to avoid overflow
4512 - update lconf's list of debug and subsystem masks
4513 - fix lfs find --obd (2510)
4514 - /proc tunable for disabling filter read caching (2591)
4515 - stop rpm packages from altering slapd.conf (2301)
4516 - disable nagle in the socknal under 0conf (2578)
4517 - choose mds inode size based on stripe count (2572)
4518 - fix kernel-source rpm problems (2516)
4520 - add --disable-doc to avoid pdf generation (2421)
4521 - update documentation, tests, type-os, comments
4522 - avoid format warnings on ia64
4523 - remove the TOE NAL
4524 - tiny code cleanups by removing unused fields
4526 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
4529 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
4530 - remove the most blatant lies from BUILDING file (2371)
4531 - change default debug level to reasonable production setting
4532 - reduce client side cache size to reduce cache flush time
4533 - reduce max RPCs in flight to avoid unnecessary file fragmentation
4534 - make TCP zerocopy and pinger support enabled by default (2476)
4535 - sync writes completed after process exits caused crashes (2319)
4536 - maintain correct mount count on the MDS (2356)
4537 - backout 1557, because 2316 wasn't really fixed
4538 - better file I/O statistics gathering in /proc
4539 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
4540 - another kernel patch to fix zero-copy TCP function export
4541 - don't take duplicate lock when processing re-sent getattr (2420)
4542 - lctl uses obd_self_export instead of creating new conn (2353)
4543 - MDS/OST recovery case which requires object creation asserted (2425)
4544 - move lfs from /usr/sbin to /usr/bin in packages
4545 - fix race between mds_client_add and mds_client_free (2417)
4546 - use kmalloc instead of slabs in portals (2430)
4547 - don't create duplicate records when a failover MDS is present (2442)
4548 - remove unnecessary mount age check (2332)
4549 - don't remove directory inodes from locks prematurely (2451)
4550 - don't break if MDS service name is the same as hostname (2103)
4551 - fix races in client write RPC generation when cache full (2482)
4553 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
4556 - remove now-unused request->rq_obd (278)
4557 - if an allocation fails, print out how much memory we've used (1933)
4558 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
4559 - add the "configurable stack size" patch to most series files (1256)
4560 - ability to write large log records, for 100+ OST configs (2306)
4561 - fix NULL deref when filter_prep fails (2314)
4562 - fix operator precedence error in filter_sync
4563 - dynamic allocation of socknal TX descriptors (2315)
4564 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
4565 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
4566 - ensure that CWARN messages in Portals always get to the syslog
4567 - __init/__exit are not for prototype decls (ldlm_init/exit)
4568 - x86-64 compile warning fixes
4569 - fix gateway LMC keyword conflict (2318)
4570 - fix MDS lock inversions in getattr/reint paths (1844)
4571 - fix a rare lock re-ordering bug, which caused deadlock (2322)
4572 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
4573 - DLM race condition prevented some lock evictions (2328)
4574 - ENOMEM detection and retry on socknal sends (2230)
4575 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
4576 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
4577 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
4578 - better error messages when a client is rejected during recovery (1505)
4579 - avoid cancelling locks which were never granted, after failure (2330)
4580 - fix i_sem/journal inversion in mds_client_add (2333)
4581 - fix truncate/getattr lock cycle deadlock (2334)
4582 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
4583 - fix two rare exit paths which could leak an l_lock() ref (2321)
4584 - fencepost error in MDS/OST orphan recovery (2226)
4585 - make log record alignment 8 bytes (1988)
4586 - lstripe now fails when requested offset > ost_count (2237)
4587 - ensure that all kernel series have a complete list.h (1607)
4588 - fix crashes in special-file operations (2316)
4589 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
4591 - allow configurable automake binary, for testing new versions
4592 - small update to the lfs documentation
4594 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
4596 * fix negative export reference count in fsfilt_sync (2312)
4598 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
4599 * release candidate 0.9.1
4601 - orphans are moved into the PENDING directory for possible recovery
4602 - replayed opens now open by fid for orphan/rename safety (1042)
4603 - last close of an orphan inode generates a transno (683)
4604 - chdir() and mount() now pin the directory entry (1020)
4605 - avoid CERROR in normal ll_setattr_raw() error case (1500)
4606 - discard very old requests without processing them (1502)
4607 - remove some common, well-understood CERRORs (1505)
4608 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
4609 - clear "grant" flags in OST replies until OST grant code lands (1644)
4610 - fix read performance by not clobbering i_blksize on client (1598)
4611 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
4612 - make LRU size a /proc tunable, clears locks when reduced (707)
4613 - fix some lprocfs rot that prevented ptlbd from loading (1732)
4614 - server locks take references on exports now (1558)
4615 - build fixes for 2.4.20-rh trees (1663)
4616 - return an error from lov_create if all OSCs are inactive (1751)
4617 - fix import levels when a reconnect happens without a timeout (1597)
4618 - exit early from mds_open if we get a lookup error (1749)
4619 - partial page read at EOF wouldn't wait for disk before sending (1642)
4620 - avoid NULL deref in obdfilter when reading page past EOF (1592)
4621 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
4622 - fix LBUG in ll_it_open_error with rc = -2 (1861)
4623 - write/truncate lock inversion (1639)
4624 - Don't auto-load obdclass, portals modules during cleanup (1495)
4625 - fix timestamps from jumping to "now" (1763)
4626 - extra journal assertions (1648)
4627 - add an extra multiunlink test (1771)
4628 - fix read_record/write_record API (1776)
4629 - fix leak of offset_extent, possible incorrect i_size later (1772)
4630 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
4631 - Debugging for the kqswnal_get_idle_tx problems (1820)
4632 - Allow recovery to be attempted multiple times (1536)
4633 - Write out MDS last_rcvd file after it is first created (1600)
4634 - Fix tx_descriptor leak in failed transmit situations (1827)
4635 - ext3 journaling fixes for assertion failure after IO error (1871)
4636 - class_export_put() on freed export after completion AST error (1896)
4637 - Fix revalidate looping in VFS (1322)
4638 - Don't access a freed export during MDS_REINT timeout (1521)
4639 - Add open-unlink recovery support on the MDS (1673,1764)
4640 - Return an error if no MDS data was read from last_rcvd (1946)
4641 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
4642 - Don't setup a disk device that doesn't match exported UUID (317)
4643 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
4644 - avoid committing NULL handle in force close
4645 - local.sh is now a one-stripe LOV configuration
4646 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
4647 - fix ext3 htree duplicate directory entry corruption (1516)
4648 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
4649 - update to new LOV EA format (2097)
4650 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
4652 2003-06-15 Phil Schwan <phil@clusterfs.com>
4655 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
4656 - per-import/export recovery handling (958, 931, 959)
4657 - multiple last-rcvd slots, for serving multiple FSes (949)
4658 - connections are again shared between multiple imp/exports (963, 964)
4659 - "umount -f" would hang if any requests needed to be sent (393, 978)
4660 - avoid pinning large req buffer by copying for queued messages (989)
4661 - add "uuid" to "lctl device" command to help upcalls (991)
4662 - "open" RPCs with transnos would confuse recovery counters (1037)
4663 - do proper endian conversion of all wire messages (288, 340, 891)
4664 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
4665 - call waiting locks callback from LDLM recovery thread (1127, 1151)
4666 - fix ptlrpc_connection leak in target_handle_connect (1174)
4667 - fix import refcounting bug in OST and MDS cleanup (1134)
4668 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
4669 - fix very unlikely obd_types race condition (501)
4670 - remove osc_open hack for echo_client (1187)
4671 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
4672 - a failure in read_inode2 leads to deadlock (1139)
4673 - cancel ack-locks as soon as transaction is committed (1072)
4674 - fix major leaks and crashes in the bulk I/O path (937, 1057)
4675 - make sure to commitrw after any preprw to avoid deadlock (1162)
4676 - failing to execute a file in a lustre FS would lock inode (1203)
4677 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
4678 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
4679 - obd_uuid2tgt would walk off the end of the list (1255)
4680 - on IA64 the osc would give portals incorrect bulk size (1258)
4681 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
4682 - fix lock inversion caused by new llite matching code (1282)
4683 - limit the number of dirty pages on a client to 10MB (1286)
4684 - timed out locks were not being corrected cancelled (1289)
4685 - fix O_DIRECT above 4GB on IA-32 (1292)
4686 * major user-visible changes
4687 - fail out/fail over policy now controlled by the upcall (993)
4689 - add OBD_PING to check server availability and failure (954)
4690 - lustre messages are now sent in sending host order (288, 340, 891)
4691 - add eadatalen to MDS getattr reply (340)
4692 - OST read replies may contain second buffer, with per-page status (593)
4694 2003-03-11 Phil Schwan <phil@clusterfs.com>
4697 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
4698 - failed open()s could cause deadlock; fixed (867, 869)
4699 - stop cancelling OST locks when files are closed (481)
4700 - overlapping XID spaces caused network corruption (851, 853)
4701 - fix unsafe fsfilt counter arithmetic; change to atomic_t
4702 - setattr_raw added, to do single-RPC, server-side setattrs
4703 - lmc/lconf syntax change for OST UUIDs
4704 - fix crashy race condition between ptlrpc_free_req and osc_close
4705 - don't use request in mdc_enqueue if we hit a timeout (889)
4706 - don't set the inode i_size for regular files from the MDS (896)
4707 - handle out of order completion AST (842)
4708 - don't LBUG if a lock request times out after receiving AST (913)
4709 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
4710 - if a bad lock AST arrives, send an error instead of dropping entirely
4711 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
4712 - fix leak in bulk IO when only partially completed (899, 900, 926)
4713 - fix O_DIRECT for ia64 (55)
4714 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
4715 - C-z after timeout could hang a process forever; fixed (977)
4717 - client-side I/O cache (678, 924, 929, 941, 970)
4719 - READPAGE and SETATTRs which don't take server-side locks get
4722 2003-02-11 Phil Schwan <phil@clusterfs.com>
4725 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
4726 - Fix fsfilt_extN_readpage() to read a full page of directory
4727 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
4728 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
4729 - Fix imbalanced LOV object allocation and out-of-bound access (469)
4730 - Most intent operations were removed, in favour of a new RPC mode
4731 that does a single RPC to the server and bypasses most of the VFS
4732 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
4733 - Aggressively cancel local locks on DLM servers
4734 - mds_reint_unlink sends EA to the client if it's the last nlink.
4735 client uses that EA to unlink OST objects.
4736 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
4737 - recursive symlinks were fixed (439)
4738 - fixed NULL deref in DEBUG_REQ
4739 - filter_update_lastobjid no longer calls sync, which annoyed extN
4740 - fixed multi-client small-writes to a single file problem (445)
4741 - fixed mtime updates during file writes (607)
4742 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
4743 - fixed bug in obd_brw_read/write() (under guise of testing 367)
4744 - fixed Linux OST size reporting problem (444, 656)
4745 - OST now updates object mtime with writes or setattr (607, 619)
4746 - client verifies file size before zeroing page past EOF (445)
4747 - OST now writes last allocated objid to disk with allocation (108)
4748 - LOV on echo now works (409)
4750 - mds_reint_unlink sends a new buffer, with the EA included. this
4751 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
4752 set if a regular file was being unlinked, and it was the last link
4753 - use PtlGet from the target for bulk writes (315)
4754 - OST now updates object mtime with writes or setattr (607, 619)
4755 - LDLM now has a grant-time callback to revalidate locked items, if
4757 - Many MDS operations were reorganized to combat race conditions
4759 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
4760 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
4762 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
4765 - Fully reactivate OST imports after reconnection (512, others)
4766 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
4767 - More graceful error handling for truncating on dead OST (515)
4768 - Don't error out unless we're actually accessing dead stripes (474)
4769 - Fix garbage sizes when stripes are missing (410)
4770 - LRU counters were broken, causing constant lock purge (433, 432)
4771 - garbage on read from stripes with failed OSTs (441)
4772 - mark OSCs as active before reconnecting during recovery (438)
4773 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
4774 - lfind did not preserve OST order in output (443)
4775 - symlinks cause hung clients, incorrect data (439)
4776 - stop dereferencing request after dropping refcount (457)
4777 - don't LASSERT(spin_is_locked) on non-SMP (455)
4778 - fixes for many rename() bugs
4779 - fstat didn't correctly synchronize attributes (399)
4780 - server must handle lock cancellation during blocking AST prep (487)
4781 - bulk descriptors were free()d too soon (511)
4782 - fix paths in lconf, which would load incorrect modules (451, 507)
4783 - fix confusing lconf 'host not found' error message (386)
4784 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
4785 - fix race condition in mdc_blocking_ast() for inode access (526)
4786 - fix lov_unpackmd() unpacking wrong number of stripes (537)
4787 - fix lov_set_osc_active() marking wrong OSC inactive (440)
4788 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
4789 - fix multiple writes of stripe MD to MDS (358, maybe 519)
4790 - fix lstripe in several ways (kernel side) (527)
4791 - fix request leak in ldlm_cli_enqueue (262)
4792 - incorrect OSC was marked inactive after OST failure
4793 - call mds_fs_cleanup before unmounting filesystem (524)
4794 - fix races between taking ns_lock and ldlm_lock_change_resource
4795 - fix races updating LOV export open file list
4796 - fix lov_enqueue error path, avoid decref-ing bad lock handle
4797 - fix recovery NULL deref in ldlm_cli_cancel_unused
4798 - fix some DLM races by using new hash table for lock handles (419)
4799 - permit the client to specify desired inodes, at replay
4800 - duplicate requests when we queue them for replay reintegration
4801 - fix last_rcvd offset calculation
4802 - sync after each recovered transaction, so we always make progress
4803 - never, not always, ERESTART requests without transnos
4804 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
4806 - skip replay if the MDS says that the client is already connected
4807 - don't check for a recovery-enabled export to match lctl's UUID
4808 - don't INC_USE_COUNT for phantom exports
4809 - don't crash when cleaning up phantom exports (567)
4810 - don't double-finish or set replay data for errored mdc_open requests
4811 - abort requests when they time out, so we don't get old replies
4812 - send/receive replies for AST messages again
4813 - if the client says that it doesn't have the lock, cancel it on the
4815 - if we timeout during I/O, don't try to cancel an in-use lock; instead
4816 mark it as destroyed, it will all work out when decref is called
4817 - fix module use counts (22, 581)
4819 - ASTs now expect a reply (server cancels lock on error reply)
4821 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
4824 - fix many simultaneous client startup (392)
4825 - fix dentry->d_it clobbering
4826 - credentials weren't being shipped for readdir/getattr operations
4827 - remove invalid assertions triggered during some concurrent MD
4829 - proper Lustre versions added (336, 389)
4830 - fix memory leak for create error case (398)
4831 - fix LOV locking bug that would get cli/srv out of sync
4832 - fix echo client over LOV (409)
4833 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
4834 - fix double-O_EXCL intent crash (424)
4835 - avoid sending multiple lock CANCELs (352)
4837 - MDS can do multi-client recovery (modulo bugs in new code)
4839 - many updates, edits, cleanups
4841 2002-11-18 Phil Schwan <phil@clusterfs.com>
4844 - fix null d_it dereference (346)
4845 - fix full OST/dbench hang (333)
4846 - fix permission problem with file removal (286)
4847 - fix removal of OSCs from LOV when they fail
4848 - fix NULL deref during bulk timeout (214)
4849 - fix problems related to multiple filesystems on one MDS (241)
4850 - fixed serious subtle metadata locking bugs
4851 - free locks on clients when inodes are removed due to memory
4853 - fix inode pointer in lock data (285)
4854 - partial support for multiple MDS on a single host (241)
4855 - data locks weren't cancelled at clear_inode time (290, 311)
4856 - intent locks could lead to unbounded lock growth (205)
4857 - added a maximum lock count, an LRU list, and a flusher
4858 - fix multiple rename (365)
4859 - properly abstracted the echo client
4860 - OSC locked 1 byte too many; fixed
4861 - rewrote brw callback code:
4862 - fixed recovery bugs related to LOVs (306)
4863 - fixed too-many-pages-in-one-write crash (191)
4864 - fixed (again) crash in sync_io_timeout (214)
4865 - probably fixed callback-related race (385)
4867 - Add capability to MDS protocol
4868 - LDLM cancellations and callbacks on different portals
4870 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
4873 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
4874 - defer open object destruction to close time (601981, 138)
4875 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
4876 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
4877 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
4878 - don't call obd_setattr in ll_file_release for destroyed objects
4879 * protocol change to lustre_msg: move |version| and add |flags|
4880 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
4881 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
4882 * added replay of create, unlink, link and rename operations during
4883 MDS failover; recovery should be much more robust now
4884 * remove failed OSCs from LOVs (only lov_create uses this so far)
4885 * the lustre-HOWTO was brought (more) up to date (582544)
4887 2002-10-23 Phil Schwan <phil@clusterfs.com>
4890 - in-use dentries weren't being reused properly (617851)
4891 - prevent multiple LDLM setup (599178)
4892 - fix LOV size calculations for truncate (617853)
4893 - fix client handling of MDS intent errors (POSIX)
4894 - fix permission bug in lovstripe.c test (624321)
4895 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
4896 - truncate past end of file could corrupt data
4897 - proper cleanup after timeouts, crashes, etc (592524, 550815)
4898 - a race in recovery could return ETIMEDOUT to apps (623947)
4899 - building outside the source directory was fixed
4900 * the lustre-HOWTO was brought (more) up to date (582544)
4901 * major progress was made on recovery functionality
4903 2002-10-10 Phil Schwan <phil@clusterfs.com>
4906 - recovery deadlock fix
4907 - rm -rf causes LBUG fix (617817)
4908 - file open by multiple tasks fix (618962)
4909 - directory permissions bugs (602707 and 620007)
4910 - journal_stop fixed with locking (611313)
4911 - O_APPEND failures resolved (618273, perhaps 614459)
4912 - lconf PATH fix (619770)
4913 - IA64 build fix (621450)
4914 - RPC buffer sizes scale with amount of memory
4916 2002-10-01 Phil Schwan <phil@clusterfs.com>
4919 - locks would be cancelled without throwing away data pages,
4920 resulting in inconsistent data (605627)
4921 - inode attributes were not always being refreshed (605627, 612449)
4922 - lconf now continues to cleanup after lctl reports an error
4923 - MDS now enforces user permissions (602707)
4924 - lprocfs cleanup fixed, but not yet enabled (614157)
4925 - fixed infinite server hang, should a client not respond to an AST
4926 - avoid going into recovery if user calls readlink() with a buffer
4927 that's too small (613941)
4928 - AST RPCs no longer require replies (614867) -- this may be changed
4929 - don't crash server if client sends an IOV that's too big (611336)
4930 - fixed lock conversion deadlock (611892)
4931 - fixed the following of symlinks (614622)
4932 * recovery: the server can remove locks from a client that dies, other
4933 clients can make progress
4934 * more extN patch fixes
4935 * compile-time configurable ptlrpc buffer allocations
4937 - collaborative read cache document
4938 - Lustre Lite Performance CDR document-in-progress
4940 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4943 - fix typo in patch-2.4.18
4945 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4948 - clear ptlrpc request each time in handle_incoming_request()
4949 - unlink of files now destroys the object on the OST
4951 2002-09-19 Peter Braam <braam@clusterfs.com>
4953 * add hard link support
4954 * change obdfile creation method
4955 * kernel patch changed
4957 2002-09-19 Peter Braam <braam@clusterfs.com>
4960 - stack overflow bug in extN fixed
4962 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
4964 * documentation updates
4965 - add man pages for config tools
4966 - update tests/README to describe testing with new config tools
4967 - finish metadata API descriptions
4968 * bug fixes and cleanups
4969 - statfs workaround for 16TB limit
4970 - LOV stripe allocation improved, can stripe on subset of OSTs
4971 - LOV file size/IO offset was wrong for files > 4GB in size
4972 - object EA data was being dropped, caused files to be unreadable
4973 - memory overflow with non-LOV OST caused memory corruption
4974 - fixed regression tests to work with new config tools, obdfilter
4975 - fixed bug when directory size became larger than 1 block
4976 - fixed bug (for single client case) when PWD was deleted
4977 - invalidate local directory pages when doing intent-based ops
4978 - avoid LDLM oops when lock callback contained bad data
4980 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
4982 * documentation updates
4983 * bug fixes and cleanups
4984 - configuration tools
4987 - 64-bit compile warnings
4988 - 64-bit internal statfs data
4990 * test_brw on persistent OST devices
4992 * lprocfs (disabled)
4994 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
4996 * documentation updates
4997 * bug fixes and cleanups
4998 * configuration tools
5000 2002-08-30 Peter J. Braam <braam@clusterfs.com>
5003 * many small fixes to 0_5_4
5004 * io/network handling
5005 * thinkos in MDS operations
5007 2002-08-24 Peter J. Braam <braam@clusterfs.com>
5010 * crucial basic fixes to 0.5.3
5011 * IOR, Iozone work over Elan
5014 2002-08-07 Phil Schwan <phil@clusterfs.com>
5015 * version 0_5_3, our first alpha
5016 * we use the new Portals iovs
5017 * documentation updates
5018 * bug fixes and cleanups
5019 * small changes in the DLM wire protocol
5021 2002-07-25 Peter J. Braam <braam@clusterfs.com>
5022 * version 0_5_1 with some initial stability,
5023 * locking on MD and file I/O.
5024 * documentation updates
5025 * several bug fixes since 0.5.0
5026 * small changes in wire protocol
5028 2002-07-18 Phil Schwan <phil@clusterfs.com>
5030 * delivered as Lustre Light Alpha
5031 * fixed a crash after handling invalid MDS requests
5032 * fixed directory pages for architectures with non-4k pages sizes
5034 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
5035 * release version v0_4_4
5036 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
5038 * Fixes a number of interruption problems with OST operations.
5039 * Update documentation for portals header changes
5040 * Move all wire protocol structs/defines to lustre_idl.h
5041 * Fixes symlink length bug.
5042 * Add tcpdump to repository.
5044 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
5045 * release version v0_4_3
5046 * Fixes statfs for inodes on extN.
5047 * Fixes bug in runtests which would delete /etc/hosts.
5048 * Use 64-bit object IDs wherever possible (not into VFS though)
5049 Remove ost_get_info, which is unused by lustre, and out of date.
5051 2002-07-03 Peter Braam <braam@clusterfs.com>
5052 * release version v0_4_2 Fixes a lookup error (type not passed)
5053 * move forward to head of Portals
5054 * move forward to latest Lustre kernel
5056 2002-06-25 Peter Braam <braam@clusterfs.com>
5057 * release version v0_4_1. Hopefully stable on single node use.