1 tbd Sun Microsystems, Inc.
3 * Support for networks:
4 socklnd - any kernel supported by Lustre,
5 qswlnd - Qsnet kernel modules 5.20 and later,
6 openiblnd - IbGold 1.8.2,
7 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, 1.3, and 1.4.1
8 viblnd - Voltaire ibhost 3.4.5 and later,
9 ciblnd - Topspin 3.2.0,
10 iiblnd - Infiniserv 3.3 + PathBits patch,
11 gmlnd - GM 2.1.22 and later,
12 mxlnd - MX 1.2.10 or later,
13 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
17 Description: should update lp_alive for non-router peers.
19 Severity : enhancement
21 Description: LNet router shuffler.
23 Severity : enhancement
25 Description: LNet fine grain routing support.
29 Description: router checker stops working when system wall clock goes backward
30 Details : use monotonic timing source instead of system wall clock time.
32 Severity : enhancement
34 Description: avoid asymmetrical router failures
36 Severity : enhancement
38 Description: multiple-instance support for kptllnd
42 Description: ksocknal_close_conn_locked connection race
43 Details : A race was possible when ksocknal_create_conn calls
44 ksocknal_close_conn_locked for already closed conn.
46 Severity : enhancement
48 Description: port router pinger to userspace
52 Description: kptllnd HELLO protocol deadlock
53 Details : kptllnd HELLO protocol doesn't run to completion in finite time
57 Description: LNet selftest fixes and enhancements
59 Severity : enhancement
61 Description: allow a test node to be a member of multiple test groups
63 Severity : enhancement
65 Description: MXLND: eliminate hosts file, use arp for peer nic_id resolution
66 Details : an update from the upstream developer Scott Atchley.
69 -------------------------------------------------------------------------------
70 2009-07-31 Sun Microsystems, Inc.
72 * Support for networks:
73 socklnd - any kernel supported by Lustre,
74 qswlnd - Qsnet kernel modules 5.20 and later,
75 openiblnd - IbGold 1.8.2,
76 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, 1.3, and 1.4.1
77 viblnd - Voltaire ibhost 3.4.5 and later,
78 ciblnd - Topspin 3.2.0,
79 iiblnd - Infiniserv 3.3 + PathBits patch,
80 gmlnd - GM 2.1.22 and later,
81 mxlnd - MX 1.2.1 or later,
82 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
86 Description: router_proc.c is rewritten to use sysctl-interface for parameters
87 residing in /proc/sys/lnet
89 Severity : enhancement
91 Description: add a new LND optiion to control peer buffer credits on routers
95 Description: Fixing deadlock in usocklnd
96 Details : A deadlock was possible in usocklnd due to race condition while
97 tearing connection down. The problem resulted from erroneous
98 assumption that lnet_finalize() could have been called holding
102 Bugzilla : 13621, 15983
103 Description: Protocol V2 of o2iblnd
104 Details : o2iblnd V2 has several new features:
105 . map-on-demand: map-on-demand is disabled by default, it can
106 be enabled by using modparam "map_on_demand=@value@", @value@
107 should >= 0 and < 256, 0 will disable map-on-demand, any other
108 valid value will enable map-on-demand.
109 Oi2blnd will create FMR or physical MR for RDMA if fragments of
111 Enable map-on-demand will take less memory for new connection,
112 but a little more CPU for RDMA.
113 . iWARP : to support iWARP, please enable map-on-demand, 32 and 64
114 are recommanded value. iWARP will probably fail for value >=128.
115 . OOB NOOP message: to resolve deadlock on router.
116 . tunable peer_credits_hiw: (high water to return credits),
117 default value of peer_credits_hiw equals to (peer_credits -1),
118 user can change it between peer_credits/2 and (peer_credits - 1).
119 Lower value is recommended for high latency network.
120 . tunable message queue size: it always equals to peer_credits,
121 higher value is recommended for high latency network.
122 . It's compatible with earlier version of o2iblnd
126 Description: Fixing 'running out of ports' issue
127 Details : Add a delay before next reconnect attempt in ksocklnd in
128 the case of lost race. Limit the frequency of query-requests
129 in lnet. Improved handling of 'dead peer' notifications in
134 Description: Change ptllnd timeout and watchdog timers
135 Details : Add ptltrace_on_nal_failed and bump ptllnd timeout to match
136 Portals wire timeout.
140 Description: One down Lustre FS hangs ALL mounted Lustre filesystems
141 Details : Shared routing enhancements - peer health detection.
145 Description: IB path MTU mistakenly set to 1st path MTU when ib_mtu is off
146 Details : See comment 46 in bug 11245 for details - it's indeed a bug
147 introduced by the original 11245 fix.
151 Description: uptllnd credit overflow fix
152 Details : kptl_msg_t::ptlm_credits could be overflown by uptllnd since
157 Description: socklnd protocol version 3
158 Details : With current protocol V2, connections on router can be
159 blocked and can't receive any incoming messages when there is no
160 more router buffer, so ZC-ACK can't be handled (LNet message
161 can't be finalized) and will cause deadlock on router.
162 Protocol V3 has a dedicated connection for emergency messages
163 like ZC-ACK to router, messages on this dedicated connection
164 don't need any credit so will never be blocked. Also, V3 can send
165 keepalive ping in specified period for router healthy checking.
167 -------------------------------------------------------------------------------
168 12-31-2008 Sun Microsystems, Inc.
170 * Support for networks:
171 socklnd - any kernel supported by Lustre,
172 qswlnd - Qsnet kernel modules 5.20 and later,
173 openiblnd - IbGold 1.8.2,
174 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
175 viblnd - Voltaire ibhost 3.4.5 and later,
176 ciblnd - Topspin 3.2.0,
177 iiblnd - Infiniserv 3.3 + PathBits patch,
178 gmlnd - GM 2.1.22 and later,
179 mxlnd - MX 1.2.1 or later,
180 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
189 Description: workaround for OOM from o2iblnd
190 Details : OFED needs allocate big chunk of memory for QP while creating
191 connection for o2iblnd, OOM can happen if no such a contiguous
193 QP size is decided by concurrent_sends and max_fragments of
194 o2iblnd, now we permit user to specify smaller value for
195 concurrent_sends of o2iblnd(i.e: concurrent_sends=7), which
196 will decrease memory block size required by creating QP.
200 Description: Support Zerocopy receive of Chelsio device
201 Details : Chelsio driver can support zerocopy for iov[1] if it's
202 contiguous and large enough.
206 Description: fix credit flow deadlock in uptllnd
210 Description: finalize network operation in reasonable time
211 Details : conf-sanity test_32a couldn't stop ost and mds because it
212 tried to access non-existent peer and tcp connect took
213 quite long before timing out.
217 Description: Continuous recovery on 33 of 413 nodes after lustre oss failure
218 Details : Lost reference on conn prevents peer from being destroyed, which
219 could prevent new peer creation if peer count has reached upper
224 Description: LNET Selftest results in Soft lockup on OSS CPU
225 Details : only hits when 8 or more o2ib clients involved and a session is
226 torn down with 'lst end_session' without preceeding 'lst stop'.
230 Description: concurrent_sends in IB LNDs should not be changeable at run time
231 Details : concurrent_sends in IB LNDs should not be changeable at run time
235 Description: ptl_send_rpc hits LASSERT when ptl_send_buf fails
236 Details : only hits under out-of-memory situations
239 -------------------------------------------------------------------------------
241 2009-02-07 Sun Microsystems, Inc.
243 * Support for networks:
244 socklnd - any kernel supported by Lustre,
245 qswlnd - Qsnet kernel modules 5.20 and later,
246 openiblnd - IbGold 1.8.2,
247 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
248 viblnd - Voltaire ibhost 3.4.5 and later,
249 ciblnd - Topspin 3.2.0,
250 iiblnd - Infiniserv 3.3 + PathBits patch,
251 gmlnd - GM 2.1.22 and later,
252 mxlnd - MX 1.2.1 or later,
253 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
257 Description: workaround for OOM from o2iblnd
258 Details : OFED needs allocate big chunk of memory for QP while creating
259 connection for o2iblnd, OOM can happen if no such a contiguous
261 QP size is decided by concurrent_sends and max_fragments of
262 o2iblnd, now we permit user to specify smaller value for
263 concurrent_sends of o2iblnd(i.e: concurrent_sends=7), which
264 will decrease memory block size required by creating QP.
268 Description: Support Zerocopy receive of Chelsio device
269 Details : Chelsio driver can support zerocopy for iov[1] if it's
270 contiguous and large enough.
274 Description: fix credit flow deadlock in uptllnd
278 Description: finalize network operation in reasonable time
279 Details : conf-sanity test_32a couldn't stop ost and mds because it
280 tried to access non-existent peer and tcp connect took
281 quite long before timing out.
285 Description: Continuous recovery on 33 of 413 nodes after lustre oss failure
286 Details : Lost reference on conn prevents peer from being destroyed, which
287 could prevent new peer creation if peer count has reached upper
292 Description: LNET Selftest results in Soft lockup on OSS CPU
293 Details : only hits when 8 or more o2ib clients involved and a session is
294 torn down with 'lst end_session' without preceeding 'lst stop'.
298 Description: concurrent_sends in IB LNDs should not be changeable at run time
299 Details : concurrent_sends in IB LNDs should not be changeable at run time
301 -------------------------------------------------------------------------------
303 11-03-2008 Sun Microsystems, Inc.
305 * Support for networks:
306 socklnd - any kernel supported by Lustre,
307 qswlnd - Qsnet kernel modules 5.20 and later,
308 openiblnd - IbGold 1.8.2,
309 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
310 viblnd - Voltaire ibhost 3.4.5 and later,
311 ciblnd - Topspin 3.2.0,
312 iiblnd - Infiniserv 3.3 + PathBits patch,
313 gmlnd - GM 2.1.22 and later,
314 mxlnd - MX 1.2.1 or later,
315 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
319 Description: ptl_send_rpc hits LASSERT when ptl_send_buf fails
320 Details : only hits under out-of-memory situations
322 -------------------------------------------------------------------------------
325 04-26-2008 Sun Microsystems, Inc.
327 * Support for networks:
328 socklnd - any kernel supported by Lustre,
329 qswlnd - Qsnet kernel modules 5.20 and later,
330 openiblnd - IbGold 1.8.2,
331 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5
332 viblnd - Voltaire ibhost 3.4.5 and later,
333 ciblnd - Topspin 3.2.0,
334 iiblnd - Infiniserv 3.3 + PathBits patch,
335 gmlnd - GM 2.1.22 and later,
336 mxlnd - MX 1.2.1 or later,
337 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
341 Description: excessive debug information removed
342 Details : excessive debug information removed
346 Description: ksocknal_create_conn() hit ASSERTION during connection race
347 Details : ksocknal_create_conn() hit ASSERTION during connection race
351 Description: ksocknal_send_hello() hit ASSERTION while connecting race
352 Details : ksocknal_send_hello() hit ASSERTION while connecting race
356 Description: o2iblnd/ptllnd credit deadlock in a routed config.
357 Details : o2iblnd/ptllnd credit deadlock in a routed config.
361 Description: High load after starting lnet
362 Details : gmlnd should sleep in rx thread in interruptible way. Otherwise,
363 uptime utility reports high load that looks confusingly.
367 Description: ksocklnd fails to establish connection if accept_port is high
368 Details : PID remapping must not be done for active (outgoing) connections
370 --------------------------------------------------------------------------------
372 2008-01-11 Sun Microsystems, Inc.
374 * Support for networks:
375 socklnd - any kernel supported by Lustre,
376 qswlnd - Qsnet kernel modules 5.20 and later,
377 openiblnd - IbGold 1.8.2,
378 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5
379 viblnd - Voltaire ibhost 3.4.5 and later,
380 ciblnd - Topspin 3.2.0,
381 iiblnd - Infiniserv 3.3 + PathBits patch,
382 gmlnd - GM 2.1.22 and later,
383 mxlnd - MX 1.2.1 or later,
384 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
387 Description: liblustre network error
388 Details : liblustre clients should understand LNET_ACCEPT_PORT environment
389 variable even if they don't start lnet acceptor.
393 Description: Strange message from lnet (Ignoring prediction from the future)
394 Details : Incorrect calculation of peer's last_alive value in ksocklnd
396 --------------------------------------------------------------------------------
398 2007-12-07 Cluster File Systems, Inc. <info@clusterfs.com>
400 * Support for networks:
401 socklnd - any kernel supported by Lustre,
402 qswlnd - Qsnet kernel modules 5.20 and later,
403 openiblnd - IbGold 1.8.2,
404 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5.
405 viblnd - Voltaire ibhost 3.4.5 and later,
406 ciblnd - Topspin 3.2.0,
407 iiblnd - Infiniserv 3.3 + PathBits patch,
408 gmlnd - GM 2.1.22 and later,
409 mxlnd - MX 1.2.1 or later,
410 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
414 Description: ASSERTION(me == md->md_me) failed in lnet_match_md()
418 Description: increase send queue size for ciblnd/openiblnd
422 Description: new userspace socklnd
423 Details : Old userspace tcpnal that resided in lnet/ulnds/socklnd replaced
424 with new one - usocklnd.
426 Severity : enhancement
428 Description: Console message flood
429 Details : Make cdls ratelimiting more tunable by adding several tunable in
430 procfs /proc/sys/lnet/console_{min,max}_delay_centisecs and
431 /proc/sys/lnet/console_backoff.
433 --------------------------------------------------------------------------------
435 2007-09-27 Cluster File Systems, Inc. <info@clusterfs.com>
437 * Support for networks:
438 socklnd - any kernel supported by Lustre,
439 qswlnd - Qsnet kernel modules 5.20 and later,
440 openiblnd - IbGold 1.8.2,
441 o2iblnd - OFED 1.1 and 1.2,
442 viblnd - Voltaire ibhost 3.4.5 and later,
443 ciblnd - Topspin 3.2.0,
444 iiblnd - Infiniserv 3.3 + PathBits patch,
445 gmlnd - GM 2.1.22 and later,
446 mxlnd - MX 1.2.1 or later,
447 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
451 Description: /proc/sys/lnet has non-sysctl entries
452 Details : Updating dump_kernel/daemon_file/debug_mb to use sysctl variables
456 Description: TOE Kernel panic by ksocklnd
457 Details : offloaded sockets provide their own implementation of sendpage,
458 can't call tcp_sendpage() directly
462 Description: kibnal_shutdown() doesn't finish; lconf --cleanup hangs
463 Details : races between lnd_shutdown and peer creation prevent
464 lnd_shutdown from finishing.
468 Description: open files rlimit 1024 reached while liblustre testing
469 Details : ulnds/socklnd must close open socket after unsuccessful
474 Description: build error
475 Details : fix typos in gmlnd, ptllnd and viblnd
477 ------------------------------------------------------------------------------
479 2007-07-30 Cluster File Systems, Inc. <info@clusterfs.com>
481 * Support for networks:
482 socklnd - kernels up to 2.6.16,
483 qswlnd - Qsnet kernel modules 5.20 and later,
484 openiblnd - IbGold 1.8.2,
485 o2iblnd - OFED 1.1 and 1.2
486 viblnd - Voltaire ibhost 3.4.5 and later,
487 ciblnd - Topspin 3.2.0,
488 iiblnd - Infiniserv 3.3 + PathBits patch,
489 gmlnd - GM 2.1.22 and later,
490 mxlnd - MX 1.2.1 or later,
491 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
493 2007-06-21 Cluster File Systems, Inc. <info@clusterfs.com>
495 * Support for networks:
496 socklnd - kernels up to 2.6.16,
497 qswlnd - Qsnet kernel modules 5.20 and later,
498 openiblnd - IbGold 1.8.2,
500 viblnd - Voltaire ibhost 3.4.5 and later,
501 ciblnd - Topspin 3.2.0,
502 iiblnd - Infiniserv 3.3 + PathBits patch,
503 gmlnd - GM 2.1.22 and later,
504 mxlnd - MX 1.2.1 or later,
505 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
509 Description: Initialize cpumask before use
513 Description: ASSERTION failures when upgrading to the patchless zero-copy
515 Details : This bug affects "rolling upgrades", causing an inconsistent
516 protocol version negotiation and subsequent assertion failure
517 during rolling upgrades after the first wave of upgrades.
521 Details : Change "dropped message" CERRORs to D_NETERROR so they are
522 logged instead of creating "console chatter" when a lustre
523 timeout races with normal RPC completion.
526 Details : lnet_clear_peer_table can wait forever if user forgets to
530 Details : libcfs_id2str should check pid against LNET_PID_ANY.
534 Description: added LNET self test
535 Details : landing b_self_test
540 Description: cfs_duration_{u,n}sec() wrongly calculate nanosecond part of
542 Details : do_div() macro is used incorrectly.
544 2007-04-23 Cluster File Systems, Inc. <info@clusterfs.com>
548 Description: make panic on lbug configurable
552 Description: Add OFED1.2 support to o2iblnd
553 Details : o2iblnd depends on OFED's modules, if out-tree OFED's modules
554 are installed (other than kernel's in-tree infiniband), there
555 could be some problem while insmod o2iblnd (mismatch CRC of
557 If extra Module.symvers is supported in kernel (i.e, 2.6.17),
558 this link provides solution:
559 https://bugs.openfabrics.org/show_bug.cgi?id=355
560 if extra Module.symvers is not supported in kernel, we will
561 have to run the script in bug 12316 to update
562 $LINUX/module.symvers before building o2iblnd.
563 More details about this are in bug 12316.
565 ------------------------------------------------------------------------------
567 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
568 * version 1.4.10 / 1.6.0
569 * Support for networks:
570 socklnd - kernels up to 2.6.16,
571 qswlnd - Qsnet kernel modules 5.20 and later,
572 openiblnd - IbGold 1.8.2,
574 viblnd - Voltaire ibhost 3.4.5 and later,
575 ciblnd - Topspin 3.2.0,
576 iiblnd - Infiniserv 3.3 + PathBits patch,
577 gmlnd - GM 2.1.22 and later,
578 mxlnd - MX 1.2.1 or later,
579 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
583 Description: Ptllnd didn't init kptllnd_data.kptl_idle_txs before it could be
584 possibly accessed in kptllnd_shutdown. Ptllnd should init
585 kptllnd_data.kptl_ptlid2str_lock before calling kptllnd_ptlid2str.
589 Description: gmlnd ignored some transmit errors when finalizing lnet messages.
593 Description: ptllnd logs a piece of incorrect debug info in kptllnd_peer_handle_hello.
597 Description: the_lnet.ln_finalizing was not set when the current thread is
598 about to complete messages. It only affects multi-threaded
604 Description: Changed the default kqswlnd ntxmsg=512
609 Description: Assertion failure in kernel ptllnd caused by posting passive
610 bulk buffers before connection establishment complete.
615 Description: A race in kernel ptllnd between deleting a peer and posting
616 new communications for it could hang communications -
617 manifesting as "Unexpectedly long timeout" messages.
622 Description: Kernel ptllnd lock ordering issue could hang a node.
627 Description: node crash on socket teardown race
630 Frequency : 'lctl peer_list' issued on a mx net
632 Description: Enable lctl's peer_list for MXLND
635 Frequency : after Ptllnd timeouts and portals congestion
637 Description: Credit overflows
638 Details : This was a bug in ptllnd connection establishment. The fix
639 implements better peer stamps to disambiguate connection
640 establishment and ensure both peers enter the credit flow
641 state machine consistently.
646 Description: kptllnd didn't propagate some network errors up to LNET
647 Details : This bug was spotted while investigating 11394. The fix
648 ensures network errors on sends and bulk transfers are
649 propagated to LNET/lustre correctly.
651 Severity : enhancement
653 Description: Fixed console chatter in case of -ETIMEDOUT.
655 Severity : enhancement
657 Description: Added D_NETTRACE for recording network packet history
658 (initially only for ptllnd). Also a separate userspace
659 ptllnd facility to gather history which should really be
660 covered by D_NETTRACE too, if only CDEBUG recorded history in
666 Description: o2iblnd handle early RDMA_CM_EVENT_DISCONNECTED.
667 Details : If the fabric is lossy, an RDMA_CM_EVENT_DISCONNECTED
668 callback can occur before a connection has actually been
669 established. This caused an assertion failure previously.
671 Severity : enhancement
673 Description: Multiple instances for o2iblnd
674 Details : Allow multiple instances of o2iblnd to enable networking over
675 multiple HCAs and routing between them.
679 Description: lnet deadlock in router_checker
680 Details : turned ksnd_connd_lock, ksnd_reaper_lock, and ksock_net_t:ksnd_lock
681 into BH locks to eliminate potential deadlock caused by
682 ksocknal_data_ready() preempting code holding these locks.
686 Description: Millions of failed socklnd connection attempts cause a very slow FS
687 Details : added a new route flag ksnr_scheduled to distinguish from
688 ksnr_connecting, so that a peer connection request is only turned
689 down for race concerns when an active connection to the same peer
690 is under progress (instead of just being scheduled).
692 ------------------------------------------------------------------------------
694 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
696 * Support for networks:
697 socklnd - kernels up to 2.6.16
698 qswlnd - Qsnet kernel modules 5.20 and later
699 openiblnd - IbGold 1.8.2
701 viblnd - Voltaire ibhost 3.4.5 and later
702 ciblnd - Topspin 3.2.0
703 iiblnd - Infiniserv 3.3 + PathBits patch
704 gmlnd - GM 2.1.22 and later
705 mxlnd - MX 1.2.1 or later
706 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
709 Severity : major on XT3
711 Description: libcfs overwrites /proc/sys/portals
712 Details : libcfs created a symlink from /proc/sys/portals to
713 /proc/sys/lnet for backwards compatibility. This is no
714 longer required and makes the Cray portals /proc variables
719 Description: OFED FMR API change
720 Details : This changes parameter usage to reflect a change in
721 ib_fmr_pool_map_phys() between OFED 1.0 and OFED 1.1. Note
722 that FMR support is only used in experimental versions of the
723 o2iblnd - this change does not affect standard usage at all.
725 Severity : enhancement
727 Description: new ko2iblnd module parameter: ib_mtu
728 Details : the default IB MTU of 2048 performs badly on 23108 Tavor
729 HCAs. You can avoid this problem by setting the MTU to 1024
730 using this module parameter.
732 Severity : enhancement
733 Bugzilla : 11118/11620
734 Description: ptllnd small request message buffer alignment fix
735 Details : Set the PTL_MD_LOCAL_ALIGN8 option on small message receives.
736 Round up small message size on sends in case this option
737 is not supported. 11620 was a defect in the initial
738 implementation which effectively asserted all peers had to be
739 running the correct protocol version which was fixed by always
740 NAK-ing such requests and handling any misalignments they
745 Description: When kib(nal|lnd)_del_peer() is called upon a peer whose
746 ibp_tx_queue is not empty, kib(nal|lnd)_destroy_peer()'s
747 'LASSERT(list_empty(&peer->ibp_tx_queue))' will fail.
749 Severity : enhancement
751 Description: Patchless ZC(zero copy) socklnd
752 Details : New protocol for socklnd, socklnd can support zero copy without
753 kernel patch, it's compatible with old socklnd. Checksum is
754 moved from tunables to modparams.
758 Description: When ksocknal_del_peer() is called upon a peer whose
759 ksnp_tx_queue is not empty, ksocknal_destroy_peer()'s
760 'LASSERT(list_empty(&peer->ksnp_tx_queue))' will fail.
763 Frequency : when ptlrpc is under heavy use and runs out of request buffer
765 Description: In lnet_match_blocked_msg(), md can be used without holding a
769 Frequency : very rarely
771 Description: If ksocknal_lib_setup_sock() fails, a ref on peer is lost.
772 If connd connects a route which has been closed by
773 ksocknal_shutdown(), ksocknal_create_routes() may create new
774 routes which hold references on the peer, causing shutdown
775 process to wait for peer to disappear forever.
777 Severity : enhancement
779 Description: Dump XT3 portals traces on kptllnd timeout
780 Details : Set the kptllnd module parameter "ptltrace_on_timeout=1" to
781 dump Cray portals debug traces to a file. The kptllnd module
782 parameter "ptltrace_basename", default "/tmp/lnet-ptltrace",
783 is the basename of the dump file.
786 Frequency : infrequent
788 Description: kernel ptllnd fix bug in connection re-establishment
789 Details : Kernel ptllnd could produce protocol errors e.g. illegal
790 matchbits and/or violate the credit flow protocol when trying
791 to re-establish a connection with a peer after an error or
794 Severity : enhancement
796 Description: Allow /proc/sys/lnet/debug to be set symbolically
797 Details : Allow debug and subsystem debug values to be read/set by name
798 in addition to numerically, for ease of use.
801 Frequency : only in configurations with LNET routers
803 Description: routes automatically marked down and recovered
804 Details : In configurations with LNET routers if a router fails routers
805 now actively try to recover routes that are down, unless they
806 are marked down by an administrator.
808 ------------------------------------------------------------------------------
810 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
813 Frequency : very rarely, in configurations with LNET routers and TCP
815 Description: incorrect data written to files on OSTs
816 Details : In certain high-load conditions incorrect data may be written
817 to files on the OST when using TCP networks.
819 ------------------------------------------------------------------------------
821 2006-07-31 Cluster File Systems, Inc. <info@clusterfs.com>
823 - rework CDEBUG messages rate-limiting mechanism b=10375
824 - add per-socket tunables for socklnd if the kernel is patched b=10327
826 ------------------------------------------------------------------------------
828 2006-02-15 Cluster File Systems, Inc. <info@clusterfs.com>
830 - fix use of portals/lnet pid to avoid dropping RPCs b=10074
831 - iiblnd wasn't mapping all memory, resulting in comms errors b=9776
832 - quiet LNET startup LNI message for liblustre b=10128
833 - Better console error messages if 'ip2nets' can't match an IP address
834 - Fixed overflow/use-before-set bugs in linux-time.h
835 - Fixed ptllnd bug that wasn't initialising rx descriptors completely
836 - LNET teardown failed an assertion about the route table being empty
837 - Fixed a crash in LNetEQPoll(<invalid handle>)
838 - Future protocol compatibility work (b_rls146_lnetprotovrsn)
839 - improve debug message for liblustre/Catamount nodes (b=10116)
841 2005-10-10 Cluster File Systems, Inc. <info@clusterfs.com>
842 * Configuration change for the XT3
843 The PTLLND is now used to run Lustre over Portals on the XT3.
844 The configure option(s) --with-cray-portals are no longer
845 used. Rather --with-portals=<path-to-portals-includes> is
846 used to enable building on the XT3. In addition to enable
847 XT3 specific features the option --enable-cray-xt3 must be
850 2005-10-10 Cluster File Systems, Inc. <info@clusterfs.com>
851 * Portals has been removed, replaced by LNET.
852 LNET is new networking infrastructure for Lustre, it includes a
853 reorganized network configuration mode (see the user
854 documentation for full details) as well as support for routing
855 between different network fabrics. Lustre Networking Devices
856 (LNDS) for the supported network fabrics have also been created
857 for this new infrastructure.
859 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
864 Frequency : rare (large Voltaire clusters only)
866 Description: the default number of reserved transmit descriptors was too low
867 for some large clusters
868 Details : As a workaround, the number was increased. A proper fix includes
871 2005-06-02 Cluster File Systems, Inc. <info@clusterfs.com>
876 Frequency : occasional (large-scale events, cluster reboot, network failure)
878 Description: too many error messages on console obscure actual problem and
879 can slow down/panic server, or cause recovery to fail repeatedly
880 Details : enable rate-limiting of console error messages, and some messages
881 that were console errors now only go to the kernel log
883 Severity : enhancement
885 Description: add /proc/sys/portals/catastrophe entry which will report if
886 that node has previously LBUGged
888 2005-04-06 Cluster File Systems, Inc. <info@clusterfs.com>
890 - update gmnal to use PTL_MTU, fix module refcounting (b=5786)
892 2005-04-04 Cluster File Systems, Inc. <info@clusterfs.com>
894 - handle error return code in kranal_check_fma_rx() (5915,6054)
896 2005-02-04 Cluster File Systems, Inc. <info@clusterfs.com>
898 - update vibnal (Voltaire IB NAL)
899 - update gmnal (Myrinet NAL), gmnalid
901 2005-02-04 Eric Barton <eeb@bartonsoftware.com>
903 * Landed portals:b_port_step as follows...
905 - removed CFS_DECL_SPIN*
906 just use 'spinlock_t' and initialise with spin_lock_init()
908 - removed CFS_DECL_MUTEX*
909 just use 'struct semaphore' and initialise with init_mutex()
911 - removed CFS_DECL_RWSEM*
912 just use 'struct rw_semaphore' and initialise with init_rwsem()
914 - renamed cfs_sleep_chan -> cfs_waitq
915 cfs_sleep_link -> cfs_waitlink
917 - fixed race in linux version of arch-independent socknal
918 (the ENOMEM/EAGAIN decision).
920 - Didn't fix problems in Darwin version of arch-independent socknal
921 (resetting socket callbacks, eager ack hack, ENOMEM/EAGAIN decision)
923 - removed libcfs types from non-socknal header files (only some types
924 in the header files had been changed; the .c files hadn't been