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 1.2.1 or later,
13 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
17 Description: kptllnd HELLO protocol deadlock
18 Details : kptllnd HELLO protocol doesn't run to completion in finite time
22 Description: LNet selftest fixes and enhancements
24 Severity : enhancement
26 Description: allow a test node to be a member of multiple test groups
28 Severity : enhancement
30 Description: MXLND: eliminate hosts file, use arp for peer nic_id resolution
31 Details : an update from the upstream developer Scott Atchley.
34 -------------------------------------------------------------------------------
35 tbd Sun Microsystems, Inc.
37 * Support for networks:
38 socklnd - any kernel supported by Lustre,
39 qswlnd - Qsnet kernel modules 5.20 and later,
40 openiblnd - IbGold 1.8.2,
41 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, 1.3, and 1.4.1
42 viblnd - Voltaire ibhost 3.4.5 and later,
43 ciblnd - Topspin 3.2.0,
44 iiblnd - Infiniserv 3.3 + PathBits patch,
45 gmlnd - GM 2.1.22 and later,
46 mxlnd - MX 1.2.1 or later,
47 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
49 Severity : enhancement
51 Description: add a new LND optiion to control peer buffer credits on routers
55 Description: Fixing deadlock in usocklnd
56 Details : A deadlock was possible in usocklnd due to race condition while
57 tearing connection down. The problem resulted from erroneous
58 assumption that lnet_finalize() could have been called holding
62 Bugzilla : 13621, 15983
63 Description: Protocol V2 of o2iblnd
64 Details : o2iblnd V2 has several new features:
65 . map-on-demand: map-on-demand is disabled by default, it can
66 be enabled by using modparam "map_on_demand=@value@", @value@
67 should >= 0 and < 256, 0 will disable map-on-demand, any other
68 valid value will enable map-on-demand.
69 Oi2blnd will create FMR or physical MR for RDMA if fragments of
71 Enable map-on-demand will take less memory for new connection,
72 but a little more CPU for RDMA.
73 . iWARP : to support iWARP, please enable map-on-demand, 32 and 64
74 are recommanded value. iWARP will probably fail for value >=128.
75 . OOB NOOP message: to resolve deadlock on router.
76 . tunable peer_credits_hiw: (high water to return credits),
77 default value of peer_credits_hiw equals to (peer_credits -1),
78 user can change it between peer_credits/2 and (peer_credits - 1).
79 Lower value is recommended for high latency network.
80 . tunable message queue size: it always equals to peer_credits,
81 higher value is recommended for high latency network.
82 . It's compatible with earlier version of o2iblnd
86 Description: Fixing 'running out of ports' issue
87 Details : Add a delay before next reconnect attempt in ksocklnd in
88 the case of lost race. Limit the frequency of query-requests
89 in lnet. Improved handling of 'dead peer' notifications in
94 Description: Change ptllnd timeout and watchdog timers
95 Details : Add ptltrace_on_nal_failed and bump ptllnd timeout to match
100 Description: One down Lustre FS hangs ALL mounted Lustre filesystems
101 Details : Shared routing enhancements - peer health detection.
105 Description: IB path MTU mistakenly set to 1st path MTU when ib_mtu is off
106 Details : See comment 46 in bug 11245 for details - it's indeed a bug
107 introduced by the original 11245 fix.
111 Description: uptllnd credit overflow fix
112 Details : kptl_msg_t::ptlm_credits could be overflown by uptllnd since
117 Description: socklnd protocol version 3
118 Details : With current protocol V2, connections on router can be
119 blocked and can't receive any incoming messages when there is no
120 more router buffer, so ZC-ACK can't be handled (LNet message
121 can't be finalized) and will cause deadlock on router.
122 Protocol V3 has a dedicated connection for emergency messages
123 like ZC-ACK to router, messages on this dedicated connection
124 don't need any credit so will never be blocked. Also, V3 can send
125 keepalive ping in specified period for router healthy checking.
127 -------------------------------------------------------------------------------
128 12-31-2008 Sun Microsystems, Inc.
130 * Support for networks:
131 socklnd - any kernel supported by Lustre,
132 qswlnd - Qsnet kernel modules 5.20 and later,
133 openiblnd - IbGold 1.8.2,
134 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
135 viblnd - Voltaire ibhost 3.4.5 and later,
136 ciblnd - Topspin 3.2.0,
137 iiblnd - Infiniserv 3.3 + PathBits patch,
138 gmlnd - GM 2.1.22 and later,
139 mxlnd - MX 1.2.1 or later,
140 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
149 Description: workaround for OOM from o2iblnd
150 Details : OFED needs allocate big chunk of memory for QP while creating
151 connection for o2iblnd, OOM can happen if no such a contiguous
153 QP size is decided by concurrent_sends and max_fragments of
154 o2iblnd, now we permit user to specify smaller value for
155 concurrent_sends of o2iblnd(i.e: concurrent_sends=7), which
156 will decrease memory block size required by creating QP.
160 Description: Support Zerocopy receive of Chelsio device
161 Details : Chelsio driver can support zerocopy for iov[1] if it's
162 contiguous and large enough.
166 Description: fix credit flow deadlock in uptllnd
170 Description: finalize network operation in reasonable time
171 Details : conf-sanity test_32a couldn't stop ost and mds because it
172 tried to access non-existent peer and tcp connect took
173 quite long before timing out.
177 Description: Continuous recovery on 33 of 413 nodes after lustre oss failure
178 Details : Lost reference on conn prevents peer from being destroyed, which
179 could prevent new peer creation if peer count has reached upper
184 Description: LNET Selftest results in Soft lockup on OSS CPU
185 Details : only hits when 8 or more o2ib clients involved and a session is
186 torn down with 'lst end_session' without preceeding 'lst stop'.
190 Description: concurrent_sends in IB LNDs should not be changeable at run time
191 Details : concurrent_sends in IB LNDs should not be changeable at run time
195 Description: ptl_send_rpc hits LASSERT when ptl_send_buf fails
196 Details : only hits under out-of-memory situations
199 -------------------------------------------------------------------------------
201 2009-02-07 Sun Microsystems, Inc.
203 * Support for networks:
204 socklnd - any kernel supported by Lustre,
205 qswlnd - Qsnet kernel modules 5.20 and later,
206 openiblnd - IbGold 1.8.2,
207 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
208 viblnd - Voltaire ibhost 3.4.5 and later,
209 ciblnd - Topspin 3.2.0,
210 iiblnd - Infiniserv 3.3 + PathBits patch,
211 gmlnd - GM 2.1.22 and later,
212 mxlnd - MX 1.2.1 or later,
213 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
217 Description: workaround for OOM from o2iblnd
218 Details : OFED needs allocate big chunk of memory for QP while creating
219 connection for o2iblnd, OOM can happen if no such a contiguous
221 QP size is decided by concurrent_sends and max_fragments of
222 o2iblnd, now we permit user to specify smaller value for
223 concurrent_sends of o2iblnd(i.e: concurrent_sends=7), which
224 will decrease memory block size required by creating QP.
228 Description: Support Zerocopy receive of Chelsio device
229 Details : Chelsio driver can support zerocopy for iov[1] if it's
230 contiguous and large enough.
234 Description: fix credit flow deadlock in uptllnd
238 Description: finalize network operation in reasonable time
239 Details : conf-sanity test_32a couldn't stop ost and mds because it
240 tried to access non-existent peer and tcp connect took
241 quite long before timing out.
245 Description: Continuous recovery on 33 of 413 nodes after lustre oss failure
246 Details : Lost reference on conn prevents peer from being destroyed, which
247 could prevent new peer creation if peer count has reached upper
252 Description: LNET Selftest results in Soft lockup on OSS CPU
253 Details : only hits when 8 or more o2ib clients involved and a session is
254 torn down with 'lst end_session' without preceeding 'lst stop'.
258 Description: concurrent_sends in IB LNDs should not be changeable at run time
259 Details : concurrent_sends in IB LNDs should not be changeable at run time
261 -------------------------------------------------------------------------------
263 11-03-2008 Sun Microsystems, Inc.
265 * Support for networks:
266 socklnd - any kernel supported by Lustre,
267 qswlnd - Qsnet kernel modules 5.20 and later,
268 openiblnd - IbGold 1.8.2,
269 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
270 viblnd - Voltaire ibhost 3.4.5 and later,
271 ciblnd - Topspin 3.2.0,
272 iiblnd - Infiniserv 3.3 + PathBits patch,
273 gmlnd - GM 2.1.22 and later,
274 mxlnd - MX 1.2.1 or later,
275 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
279 Description: ptl_send_rpc hits LASSERT when ptl_send_buf fails
280 Details : only hits under out-of-memory situations
282 -------------------------------------------------------------------------------
285 04-26-2008 Sun Microsystems, Inc.
287 * Support for networks:
288 socklnd - any kernel supported by Lustre,
289 qswlnd - Qsnet kernel modules 5.20 and later,
290 openiblnd - IbGold 1.8.2,
291 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5
292 viblnd - Voltaire ibhost 3.4.5 and later,
293 ciblnd - Topspin 3.2.0,
294 iiblnd - Infiniserv 3.3 + PathBits patch,
295 gmlnd - GM 2.1.22 and later,
296 mxlnd - MX 1.2.1 or later,
297 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
301 Description: excessive debug information removed
302 Details : excessive debug information removed
306 Description: ksocknal_create_conn() hit ASSERTION during connection race
307 Details : ksocknal_create_conn() hit ASSERTION during connection race
311 Description: ksocknal_send_hello() hit ASSERTION while connecting race
312 Details : ksocknal_send_hello() hit ASSERTION while connecting race
316 Description: o2iblnd/ptllnd credit deadlock in a routed config.
317 Details : o2iblnd/ptllnd credit deadlock in a routed config.
321 Description: High load after starting lnet
322 Details : gmlnd should sleep in rx thread in interruptible way. Otherwise,
323 uptime utility reports high load that looks confusingly.
327 Description: ksocklnd fails to establish connection if accept_port is high
328 Details : PID remapping must not be done for active (outgoing) connections
330 --------------------------------------------------------------------------------
332 2008-01-11 Sun Microsystems, Inc.
334 * Support for networks:
335 socklnd - any kernel supported by Lustre,
336 qswlnd - Qsnet kernel modules 5.20 and later,
337 openiblnd - IbGold 1.8.2,
338 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5
339 viblnd - Voltaire ibhost 3.4.5 and later,
340 ciblnd - Topspin 3.2.0,
341 iiblnd - Infiniserv 3.3 + PathBits patch,
342 gmlnd - GM 2.1.22 and later,
343 mxlnd - MX 1.2.1 or later,
344 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
347 Description: liblustre network error
348 Details : liblustre clients should understand LNET_ACCEPT_PORT environment
349 variable even if they don't start lnet acceptor.
353 Description: Strange message from lnet (Ignoring prediction from the future)
354 Details : Incorrect calculation of peer's last_alive value in ksocklnd
356 --------------------------------------------------------------------------------
358 2007-12-07 Cluster File Systems, Inc. <info@clusterfs.com>
360 * Support for networks:
361 socklnd - any kernel supported by Lustre,
362 qswlnd - Qsnet kernel modules 5.20 and later,
363 openiblnd - IbGold 1.8.2,
364 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5.
365 viblnd - Voltaire ibhost 3.4.5 and later,
366 ciblnd - Topspin 3.2.0,
367 iiblnd - Infiniserv 3.3 + PathBits patch,
368 gmlnd - GM 2.1.22 and later,
369 mxlnd - MX 1.2.1 or later,
370 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
374 Description: ASSERTION(me == md->md_me) failed in lnet_match_md()
378 Description: increase send queue size for ciblnd/openiblnd
382 Description: new userspace socklnd
383 Details : Old userspace tcpnal that resided in lnet/ulnds/socklnd replaced
384 with new one - usocklnd.
386 Severity : enhancement
388 Description: Console message flood
389 Details : Make cdls ratelimiting more tunable by adding several tunable in
390 procfs /proc/sys/lnet/console_{min,max}_delay_centisecs and
391 /proc/sys/lnet/console_backoff.
393 --------------------------------------------------------------------------------
395 2007-09-27 Cluster File Systems, Inc. <info@clusterfs.com>
397 * Support for networks:
398 socklnd - any kernel supported by Lustre,
399 qswlnd - Qsnet kernel modules 5.20 and later,
400 openiblnd - IbGold 1.8.2,
401 o2iblnd - OFED 1.1 and 1.2,
402 viblnd - Voltaire ibhost 3.4.5 and later,
403 ciblnd - Topspin 3.2.0,
404 iiblnd - Infiniserv 3.3 + PathBits patch,
405 gmlnd - GM 2.1.22 and later,
406 mxlnd - MX 1.2.1 or later,
407 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
411 Description: /proc/sys/lnet has non-sysctl entries
412 Details : Updating dump_kernel/daemon_file/debug_mb to use sysctl variables
416 Description: TOE Kernel panic by ksocklnd
417 Details : offloaded sockets provide their own implementation of sendpage,
418 can't call tcp_sendpage() directly
422 Description: kibnal_shutdown() doesn't finish; lconf --cleanup hangs
423 Details : races between lnd_shutdown and peer creation prevent
424 lnd_shutdown from finishing.
428 Description: open files rlimit 1024 reached while liblustre testing
429 Details : ulnds/socklnd must close open socket after unsuccessful
434 Description: build error
435 Details : fix typos in gmlnd, ptllnd and viblnd
437 ------------------------------------------------------------------------------
439 2007-07-30 Cluster File Systems, Inc. <info@clusterfs.com>
441 * Support for networks:
442 socklnd - kernels up to 2.6.16,
443 qswlnd - Qsnet kernel modules 5.20 and later,
444 openiblnd - IbGold 1.8.2,
445 o2iblnd - OFED 1.1 and 1.2
446 viblnd - Voltaire ibhost 3.4.5 and later,
447 ciblnd - Topspin 3.2.0,
448 iiblnd - Infiniserv 3.3 + PathBits patch,
449 gmlnd - GM 2.1.22 and later,
450 mxlnd - MX 1.2.1 or later,
451 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
453 2007-06-21 Cluster File Systems, Inc. <info@clusterfs.com>
455 * Support for networks:
456 socklnd - kernels up to 2.6.16,
457 qswlnd - Qsnet kernel modules 5.20 and later,
458 openiblnd - IbGold 1.8.2,
460 viblnd - Voltaire ibhost 3.4.5 and later,
461 ciblnd - Topspin 3.2.0,
462 iiblnd - Infiniserv 3.3 + PathBits patch,
463 gmlnd - GM 2.1.22 and later,
464 mxlnd - MX 1.2.1 or later,
465 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
469 Description: Initialize cpumask before use
473 Description: ASSERTION failures when upgrading to the patchless zero-copy
475 Details : This bug affects "rolling upgrades", causing an inconsistent
476 protocol version negotiation and subsequent assertion failure
477 during rolling upgrades after the first wave of upgrades.
481 Details : Change "dropped message" CERRORs to D_NETERROR so they are
482 logged instead of creating "console chatter" when a lustre
483 timeout races with normal RPC completion.
486 Details : lnet_clear_peer_table can wait forever if user forgets to
490 Details : libcfs_id2str should check pid against LNET_PID_ANY.
494 Description: added LNET self test
495 Details : landing b_self_test
500 Description: cfs_duration_{u,n}sec() wrongly calculate nanosecond part of
502 Details : do_div() macro is used incorrectly.
504 2007-04-23 Cluster File Systems, Inc. <info@clusterfs.com>
508 Description: make panic on lbug configurable
512 Description: Add OFED1.2 support to o2iblnd
513 Details : o2iblnd depends on OFED's modules, if out-tree OFED's modules
514 are installed (other than kernel's in-tree infiniband), there
515 could be some problem while insmod o2iblnd (mismatch CRC of
517 If extra Module.symvers is supported in kernel (i.e, 2.6.17),
518 this link provides solution:
519 https://bugs.openfabrics.org/show_bug.cgi?id=355
520 if extra Module.symvers is not supported in kernel, we will
521 have to run the script in bug 12316 to update
522 $LINUX/module.symvers before building o2iblnd.
523 More details about this are in bug 12316.
525 ------------------------------------------------------------------------------
527 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
528 * version 1.4.10 / 1.6.0
529 * Support for networks:
530 socklnd - kernels up to 2.6.16,
531 qswlnd - Qsnet kernel modules 5.20 and later,
532 openiblnd - IbGold 1.8.2,
534 viblnd - Voltaire ibhost 3.4.5 and later,
535 ciblnd - Topspin 3.2.0,
536 iiblnd - Infiniserv 3.3 + PathBits patch,
537 gmlnd - GM 2.1.22 and later,
538 mxlnd - MX 1.2.1 or later,
539 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
543 Description: Ptllnd didn't init kptllnd_data.kptl_idle_txs before it could be
544 possibly accessed in kptllnd_shutdown. Ptllnd should init
545 kptllnd_data.kptl_ptlid2str_lock before calling kptllnd_ptlid2str.
549 Description: gmlnd ignored some transmit errors when finalizing lnet messages.
553 Description: ptllnd logs a piece of incorrect debug info in kptllnd_peer_handle_hello.
557 Description: the_lnet.ln_finalizing was not set when the current thread is
558 about to complete messages. It only affects multi-threaded
564 Description: Changed the default kqswlnd ntxmsg=512
569 Description: Assertion failure in kernel ptllnd caused by posting passive
570 bulk buffers before connection establishment complete.
575 Description: A race in kernel ptllnd between deleting a peer and posting
576 new communications for it could hang communications -
577 manifesting as "Unexpectedly long timeout" messages.
582 Description: Kernel ptllnd lock ordering issue could hang a node.
587 Description: node crash on socket teardown race
590 Frequency : 'lctl peer_list' issued on a mx net
592 Description: Enable lctl's peer_list for MXLND
595 Frequency : after Ptllnd timeouts and portals congestion
597 Description: Credit overflows
598 Details : This was a bug in ptllnd connection establishment. The fix
599 implements better peer stamps to disambiguate connection
600 establishment and ensure both peers enter the credit flow
601 state machine consistently.
606 Description: kptllnd didn't propagate some network errors up to LNET
607 Details : This bug was spotted while investigating 11394. The fix
608 ensures network errors on sends and bulk transfers are
609 propagated to LNET/lustre correctly.
611 Severity : enhancement
613 Description: Fixed console chatter in case of -ETIMEDOUT.
615 Severity : enhancement
617 Description: Added D_NETTRACE for recording network packet history
618 (initially only for ptllnd). Also a separate userspace
619 ptllnd facility to gather history which should really be
620 covered by D_NETTRACE too, if only CDEBUG recorded history in
626 Description: o2iblnd handle early RDMA_CM_EVENT_DISCONNECTED.
627 Details : If the fabric is lossy, an RDMA_CM_EVENT_DISCONNECTED
628 callback can occur before a connection has actually been
629 established. This caused an assertion failure previously.
631 Severity : enhancement
633 Description: Multiple instances for o2iblnd
634 Details : Allow multiple instances of o2iblnd to enable networking over
635 multiple HCAs and routing between them.
639 Description: lnet deadlock in router_checker
640 Details : turned ksnd_connd_lock, ksnd_reaper_lock, and ksock_net_t:ksnd_lock
641 into BH locks to eliminate potential deadlock caused by
642 ksocknal_data_ready() preempting code holding these locks.
646 Description: Millions of failed socklnd connection attempts cause a very slow FS
647 Details : added a new route flag ksnr_scheduled to distinguish from
648 ksnr_connecting, so that a peer connection request is only turned
649 down for race concerns when an active connection to the same peer
650 is under progress (instead of just being scheduled).
652 ------------------------------------------------------------------------------
654 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
656 * Support for networks:
657 socklnd - kernels up to 2.6.16
658 qswlnd - Qsnet kernel modules 5.20 and later
659 openiblnd - IbGold 1.8.2
661 viblnd - Voltaire ibhost 3.4.5 and later
662 ciblnd - Topspin 3.2.0
663 iiblnd - Infiniserv 3.3 + PathBits patch
664 gmlnd - GM 2.1.22 and later
665 mxlnd - MX 1.2.1 or later
666 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
669 Severity : major on XT3
671 Description: libcfs overwrites /proc/sys/portals
672 Details : libcfs created a symlink from /proc/sys/portals to
673 /proc/sys/lnet for backwards compatibility. This is no
674 longer required and makes the Cray portals /proc variables
679 Description: OFED FMR API change
680 Details : This changes parameter usage to reflect a change in
681 ib_fmr_pool_map_phys() between OFED 1.0 and OFED 1.1. Note
682 that FMR support is only used in experimental versions of the
683 o2iblnd - this change does not affect standard usage at all.
685 Severity : enhancement
687 Description: new ko2iblnd module parameter: ib_mtu
688 Details : the default IB MTU of 2048 performs badly on 23108 Tavor
689 HCAs. You can avoid this problem by setting the MTU to 1024
690 using this module parameter.
692 Severity : enhancement
693 Bugzilla : 11118/11620
694 Description: ptllnd small request message buffer alignment fix
695 Details : Set the PTL_MD_LOCAL_ALIGN8 option on small message receives.
696 Round up small message size on sends in case this option
697 is not supported. 11620 was a defect in the initial
698 implementation which effectively asserted all peers had to be
699 running the correct protocol version which was fixed by always
700 NAK-ing such requests and handling any misalignments they
705 Description: When kib(nal|lnd)_del_peer() is called upon a peer whose
706 ibp_tx_queue is not empty, kib(nal|lnd)_destroy_peer()'s
707 'LASSERT(list_empty(&peer->ibp_tx_queue))' will fail.
709 Severity : enhancement
711 Description: Patchless ZC(zero copy) socklnd
712 Details : New protocol for socklnd, socklnd can support zero copy without
713 kernel patch, it's compatible with old socklnd. Checksum is
714 moved from tunables to modparams.
718 Description: When ksocknal_del_peer() is called upon a peer whose
719 ksnp_tx_queue is not empty, ksocknal_destroy_peer()'s
720 'LASSERT(list_empty(&peer->ksnp_tx_queue))' will fail.
723 Frequency : when ptlrpc is under heavy use and runs out of request buffer
725 Description: In lnet_match_blocked_msg(), md can be used without holding a
729 Frequency : very rarely
731 Description: If ksocknal_lib_setup_sock() fails, a ref on peer is lost.
732 If connd connects a route which has been closed by
733 ksocknal_shutdown(), ksocknal_create_routes() may create new
734 routes which hold references on the peer, causing shutdown
735 process to wait for peer to disappear forever.
737 Severity : enhancement
739 Description: Dump XT3 portals traces on kptllnd timeout
740 Details : Set the kptllnd module parameter "ptltrace_on_timeout=1" to
741 dump Cray portals debug traces to a file. The kptllnd module
742 parameter "ptltrace_basename", default "/tmp/lnet-ptltrace",
743 is the basename of the dump file.
746 Frequency : infrequent
748 Description: kernel ptllnd fix bug in connection re-establishment
749 Details : Kernel ptllnd could produce protocol errors e.g. illegal
750 matchbits and/or violate the credit flow protocol when trying
751 to re-establish a connection with a peer after an error or
754 Severity : enhancement
756 Description: Allow /proc/sys/lnet/debug to be set symbolically
757 Details : Allow debug and subsystem debug values to be read/set by name
758 in addition to numerically, for ease of use.
761 Frequency : only in configurations with LNET routers
763 Description: routes automatically marked down and recovered
764 Details : In configurations with LNET routers if a router fails routers
765 now actively try to recover routes that are down, unless they
766 are marked down by an administrator.
768 ------------------------------------------------------------------------------
770 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
773 Frequency : very rarely, in configurations with LNET routers and TCP
775 Description: incorrect data written to files on OSTs
776 Details : In certain high-load conditions incorrect data may be written
777 to files on the OST when using TCP networks.
779 ------------------------------------------------------------------------------
781 2006-07-31 Cluster File Systems, Inc. <info@clusterfs.com>
783 - rework CDEBUG messages rate-limiting mechanism b=10375
784 - add per-socket tunables for socklnd if the kernel is patched b=10327
786 ------------------------------------------------------------------------------
788 2006-02-15 Cluster File Systems, Inc. <info@clusterfs.com>
790 - fix use of portals/lnet pid to avoid dropping RPCs b=10074
791 - iiblnd wasn't mapping all memory, resulting in comms errors b=9776
792 - quiet LNET startup LNI message for liblustre b=10128
793 - Better console error messages if 'ip2nets' can't match an IP address
794 - Fixed overflow/use-before-set bugs in linux-time.h
795 - Fixed ptllnd bug that wasn't initialising rx descriptors completely
796 - LNET teardown failed an assertion about the route table being empty
797 - Fixed a crash in LNetEQPoll(<invalid handle>)
798 - Future protocol compatibility work (b_rls146_lnetprotovrsn)
799 - improve debug message for liblustre/Catamount nodes (b=10116)
801 2005-10-10 Cluster File Systems, Inc. <info@clusterfs.com>
802 * Configuration change for the XT3
803 The PTLLND is now used to run Lustre over Portals on the XT3.
804 The configure option(s) --with-cray-portals are no longer
805 used. Rather --with-portals=<path-to-portals-includes> is
806 used to enable building on the XT3. In addition to enable
807 XT3 specific features the option --enable-cray-xt3 must be
810 2005-10-10 Cluster File Systems, Inc. <info@clusterfs.com>
811 * Portals has been removed, replaced by LNET.
812 LNET is new networking infrastructure for Lustre, it includes a
813 reorganized network configuration mode (see the user
814 documentation for full details) as well as support for routing
815 between different network fabrics. Lustre Networking Devices
816 (LNDS) for the supported network fabrics have also been created
817 for this new infrastructure.
819 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
824 Frequency : rare (large Voltaire clusters only)
826 Description: the default number of reserved transmit descriptors was too low
827 for some large clusters
828 Details : As a workaround, the number was increased. A proper fix includes
831 2005-06-02 Cluster File Systems, Inc. <info@clusterfs.com>
836 Frequency : occasional (large-scale events, cluster reboot, network failure)
838 Description: too many error messages on console obscure actual problem and
839 can slow down/panic server, or cause recovery to fail repeatedly
840 Details : enable rate-limiting of console error messages, and some messages
841 that were console errors now only go to the kernel log
843 Severity : enhancement
845 Description: add /proc/sys/portals/catastrophe entry which will report if
846 that node has previously LBUGged
848 2005-04-06 Cluster File Systems, Inc. <info@clusterfs.com>
850 - update gmnal to use PTL_MTU, fix module refcounting (b=5786)
852 2005-04-04 Cluster File Systems, Inc. <info@clusterfs.com>
854 - handle error return code in kranal_check_fma_rx() (5915,6054)
856 2005-02-04 Cluster File Systems, Inc. <info@clusterfs.com>
858 - update vibnal (Voltaire IB NAL)
859 - update gmnal (Myrinet NAL), gmnalid
861 2005-02-04 Eric Barton <eeb@bartonsoftware.com>
863 * Landed portals:b_port_step as follows...
865 - removed CFS_DECL_SPIN*
866 just use 'spinlock_t' and initialise with spin_lock_init()
868 - removed CFS_DECL_MUTEX*
869 just use 'struct semaphore' and initialise with init_mutex()
871 - removed CFS_DECL_RWSEM*
872 just use 'struct rw_semaphore' and initialise with init_rwsem()
874 - renamed cfs_sleep_chan -> cfs_waitq
875 cfs_sleep_link -> cfs_waitlink
877 - fixed race in linux version of arch-independent socknal
878 (the ENOMEM/EAGAIN decision).
880 - Didn't fix problems in Darwin version of arch-independent socknal
881 (resetting socket callbacks, eager ack hack, ENOMEM/EAGAIN decision)
883 - removed libcfs types from non-socknal header files (only some types
884 in the header files had been changed; the .c files hadn't been