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, and 1.3
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
22 Description: Change ptllnd timeout and watchdog timers
23 Details : Add ptltrace_on_nal_failed and bump ptllnd timeout to match
28 Description: One down Lustre FS hangs ALL mounted Lustre filesystems
29 Details : Shared routing enhancements - peer health detection.
33 Description: IB path MTU mistakenly set to 1st path MTU when ib_mtu is off
34 Details : See comment 46 in bug 11245 for details - it's indeed a bug
35 introduced by the original 11245 fix.
39 Description: uptllnd credit overflow fix
40 Details : kptl_msg_t::ptlm_credits could be overflown by uptllnd since
45 Description: socklnd protocol version 3
46 Details : With current protocol V2, connections on router can be
47 blocked and can't receive any incoming messages when there is no
48 more router buffer, so ZC-ACK can't be handled (LNet message
49 can't be finalized) and will cause deadlock on router.
50 Protocol V3 has a dedicated connection for emergency messages
51 like ZC-ACK to router, messages on this dedicated connection
52 don't need any credit so will never be blocked. Also, V3 can send
53 keepalive ping in specified period for router healthy checking.
55 -------------------------------------------------------------------------------
56 12-31-2008 Sun Microsystems, Inc.
58 * Support for networks:
59 socklnd - any kernel supported by Lustre,
60 qswlnd - Qsnet kernel modules 5.20 and later,
61 openiblnd - IbGold 1.8.2,
62 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
63 viblnd - Voltaire ibhost 3.4.5 and later,
64 ciblnd - Topspin 3.2.0,
65 iiblnd - Infiniserv 3.3 + PathBits patch,
66 gmlnd - GM 2.1.22 and later,
67 mxlnd - MX 1.2.1 or later,
68 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
77 Description: workaround for OOM from o2iblnd
78 Details : OFED needs allocate big chunk of memory for QP while creating
79 connection for o2iblnd, OOM can happen if no such a contiguous
81 QP size is decided by concurrent_sends and max_fragments of
82 o2iblnd, now we permit user to specify smaller value for
83 concurrent_sends of o2iblnd(i.e: concurrent_sends=7), which
84 will decrease memory block size required by creating QP.
88 Description: Support Zerocopy receive of Chelsio device
89 Details : Chelsio driver can support zerocopy for iov[1] if it's
90 contiguous and large enough.
94 Description: fix credit flow deadlock in uptllnd
98 Description: finalize network operation in reasonable time
99 Details : conf-sanity test_32a couldn't stop ost and mds because it
100 tried to access non-existent peer and tcp connect took
101 quite long before timing out.
105 Description: Continuous recovery on 33 of 413 nodes after lustre oss failure
106 Details : Lost reference on conn prevents peer from being destroyed, which
107 could prevent new peer creation if peer count has reached upper
112 Description: LNET Selftest results in Soft lockup on OSS CPU
113 Details : only hits when 8 or more o2ib clients involved and a session is
114 torn down with 'lst end_session' without preceeding 'lst stop'.
118 Description: concurrent_sends in IB LNDs should not be changeable at run time
119 Details : concurrent_sends in IB LNDs should not be changeable at run time
123 Description: ptl_send_rpc hits LASSERT when ptl_send_buf fails
124 Details : only hits under out-of-memory situations
127 -------------------------------------------------------------------------------
129 2009-02-07 Sun Microsystems, Inc.
131 * Support for networks:
132 socklnd - any kernel supported by Lustre,
133 qswlnd - Qsnet kernel modules 5.20 and later,
134 openiblnd - IbGold 1.8.2,
135 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
136 viblnd - Voltaire ibhost 3.4.5 and later,
137 ciblnd - Topspin 3.2.0,
138 iiblnd - Infiniserv 3.3 + PathBits patch,
139 gmlnd - GM 2.1.22 and later,
140 mxlnd - MX 1.2.1 or later,
141 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
145 Description: workaround for OOM from o2iblnd
146 Details : OFED needs allocate big chunk of memory for QP while creating
147 connection for o2iblnd, OOM can happen if no such a contiguous
149 QP size is decided by concurrent_sends and max_fragments of
150 o2iblnd, now we permit user to specify smaller value for
151 concurrent_sends of o2iblnd(i.e: concurrent_sends=7), which
152 will decrease memory block size required by creating QP.
156 Description: Support Zerocopy receive of Chelsio device
157 Details : Chelsio driver can support zerocopy for iov[1] if it's
158 contiguous and large enough.
162 Description: fix credit flow deadlock in uptllnd
166 Description: finalize network operation in reasonable time
167 Details : conf-sanity test_32a couldn't stop ost and mds because it
168 tried to access non-existent peer and tcp connect took
169 quite long before timing out.
173 Description: Continuous recovery on 33 of 413 nodes after lustre oss failure
174 Details : Lost reference on conn prevents peer from being destroyed, which
175 could prevent new peer creation if peer count has reached upper
180 Description: LNET Selftest results in Soft lockup on OSS CPU
181 Details : only hits when 8 or more o2ib clients involved and a session is
182 torn down with 'lst end_session' without preceeding 'lst stop'.
186 Description: concurrent_sends in IB LNDs should not be changeable at run time
187 Details : concurrent_sends in IB LNDs should not be changeable at run time
189 -------------------------------------------------------------------------------
191 11-03-2008 Sun Microsystems, Inc.
193 * Support for networks:
194 socklnd - any kernel supported by Lustre,
195 qswlnd - Qsnet kernel modules 5.20 and later,
196 openiblnd - IbGold 1.8.2,
197 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
198 viblnd - Voltaire ibhost 3.4.5 and later,
199 ciblnd - Topspin 3.2.0,
200 iiblnd - Infiniserv 3.3 + PathBits patch,
201 gmlnd - GM 2.1.22 and later,
202 mxlnd - MX 1.2.1 or later,
203 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
207 Description: ptl_send_rpc hits LASSERT when ptl_send_buf fails
208 Details : only hits under out-of-memory situations
210 -------------------------------------------------------------------------------
213 04-26-2008 Sun Microsystems, Inc.
215 * Support for networks:
216 socklnd - any kernel supported by Lustre,
217 qswlnd - Qsnet kernel modules 5.20 and later,
218 openiblnd - IbGold 1.8.2,
219 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5
220 viblnd - Voltaire ibhost 3.4.5 and later,
221 ciblnd - Topspin 3.2.0,
222 iiblnd - Infiniserv 3.3 + PathBits patch,
223 gmlnd - GM 2.1.22 and later,
224 mxlnd - MX 1.2.1 or later,
225 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
229 Description: excessive debug information removed
230 Details : excessive debug information removed
234 Description: ksocknal_create_conn() hit ASSERTION during connection race
235 Details : ksocknal_create_conn() hit ASSERTION during connection race
239 Description: ksocknal_send_hello() hit ASSERTION while connecting race
240 Details : ksocknal_send_hello() hit ASSERTION while connecting race
244 Description: o2iblnd/ptllnd credit deadlock in a routed config.
245 Details : o2iblnd/ptllnd credit deadlock in a routed config.
249 Description: High load after starting lnet
250 Details : gmlnd should sleep in rx thread in interruptible way. Otherwise,
251 uptime utility reports high load that looks confusingly.
255 Description: ksocklnd fails to establish connection if accept_port is high
256 Details : PID remapping must not be done for active (outgoing) connections
258 --------------------------------------------------------------------------------
260 2008-01-11 Sun Microsystems, Inc.
262 * Support for networks:
263 socklnd - any kernel supported by Lustre,
264 qswlnd - Qsnet kernel modules 5.20 and later,
265 openiblnd - IbGold 1.8.2,
266 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5
267 viblnd - Voltaire ibhost 3.4.5 and later,
268 ciblnd - Topspin 3.2.0,
269 iiblnd - Infiniserv 3.3 + PathBits patch,
270 gmlnd - GM 2.1.22 and later,
271 mxlnd - MX 1.2.1 or later,
272 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
275 Description: liblustre network error
276 Details : liblustre clients should understand LNET_ACCEPT_PORT environment
277 variable even if they don't start lnet acceptor.
281 Description: Strange message from lnet (Ignoring prediction from the future)
282 Details : Incorrect calculation of peer's last_alive value in ksocklnd
284 --------------------------------------------------------------------------------
286 2007-12-07 Cluster File Systems, Inc. <info@clusterfs.com>
288 * Support for networks:
289 socklnd - any kernel supported by Lustre,
290 qswlnd - Qsnet kernel modules 5.20 and later,
291 openiblnd - IbGold 1.8.2,
292 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5.
293 viblnd - Voltaire ibhost 3.4.5 and later,
294 ciblnd - Topspin 3.2.0,
295 iiblnd - Infiniserv 3.3 + PathBits patch,
296 gmlnd - GM 2.1.22 and later,
297 mxlnd - MX 1.2.1 or later,
298 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
302 Description: ASSERTION(me == md->md_me) failed in lnet_match_md()
306 Description: increase send queue size for ciblnd/openiblnd
310 Description: new userspace socklnd
311 Details : Old userspace tcpnal that resided in lnet/ulnds/socklnd replaced
312 with new one - usocklnd.
314 Severity : enhancement
316 Description: Console message flood
317 Details : Make cdls ratelimiting more tunable by adding several tunable in
318 procfs /proc/sys/lnet/console_{min,max}_delay_centisecs and
319 /proc/sys/lnet/console_backoff.
321 --------------------------------------------------------------------------------
323 2007-09-27 Cluster File Systems, Inc. <info@clusterfs.com>
325 * Support for networks:
326 socklnd - any kernel supported by Lustre,
327 qswlnd - Qsnet kernel modules 5.20 and later,
328 openiblnd - IbGold 1.8.2,
329 o2iblnd - OFED 1.1 and 1.2,
330 viblnd - Voltaire ibhost 3.4.5 and later,
331 ciblnd - Topspin 3.2.0,
332 iiblnd - Infiniserv 3.3 + PathBits patch,
333 gmlnd - GM 2.1.22 and later,
334 mxlnd - MX 1.2.1 or later,
335 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
339 Description: /proc/sys/lnet has non-sysctl entries
340 Details : Updating dump_kernel/daemon_file/debug_mb to use sysctl variables
344 Description: TOE Kernel panic by ksocklnd
345 Details : offloaded sockets provide their own implementation of sendpage,
346 can't call tcp_sendpage() directly
350 Description: kibnal_shutdown() doesn't finish; lconf --cleanup hangs
351 Details : races between lnd_shutdown and peer creation prevent
352 lnd_shutdown from finishing.
356 Description: open files rlimit 1024 reached while liblustre testing
357 Details : ulnds/socklnd must close open socket after unsuccessful
362 Description: build error
363 Details : fix typos in gmlnd, ptllnd and viblnd
365 ------------------------------------------------------------------------------
367 2007-07-30 Cluster File Systems, Inc. <info@clusterfs.com>
369 * Support for networks:
370 socklnd - kernels up to 2.6.16,
371 qswlnd - Qsnet kernel modules 5.20 and later,
372 openiblnd - IbGold 1.8.2,
373 o2iblnd - OFED 1.1 and 1.2
374 viblnd - Voltaire ibhost 3.4.5 and later,
375 ciblnd - Topspin 3.2.0,
376 iiblnd - Infiniserv 3.3 + PathBits patch,
377 gmlnd - GM 2.1.22 and later,
378 mxlnd - MX 1.2.1 or later,
379 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
381 2007-06-21 Cluster File Systems, Inc. <info@clusterfs.com>
383 * Support for networks:
384 socklnd - kernels up to 2.6.16,
385 qswlnd - Qsnet kernel modules 5.20 and later,
386 openiblnd - IbGold 1.8.2,
388 viblnd - Voltaire ibhost 3.4.5 and later,
389 ciblnd - Topspin 3.2.0,
390 iiblnd - Infiniserv 3.3 + PathBits patch,
391 gmlnd - GM 2.1.22 and later,
392 mxlnd - MX 1.2.1 or later,
393 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
397 Description: Initialize cpumask before use
401 Description: ASSERTION failures when upgrading to the patchless zero-copy
403 Details : This bug affects "rolling upgrades", causing an inconsistent
404 protocol version negotiation and subsequent assertion failure
405 during rolling upgrades after the first wave of upgrades.
409 Details : Change "dropped message" CERRORs to D_NETERROR so they are
410 logged instead of creating "console chatter" when a lustre
411 timeout races with normal RPC completion.
414 Details : lnet_clear_peer_table can wait forever if user forgets to
418 Details : libcfs_id2str should check pid against LNET_PID_ANY.
422 Description: added LNET self test
423 Details : landing b_self_test
428 Description: cfs_duration_{u,n}sec() wrongly calculate nanosecond part of
430 Details : do_div() macro is used incorrectly.
432 2007-04-23 Cluster File Systems, Inc. <info@clusterfs.com>
436 Description: make panic on lbug configurable
440 Description: Add OFED1.2 support to o2iblnd
441 Details : o2iblnd depends on OFED's modules, if out-tree OFED's modules
442 are installed (other than kernel's in-tree infiniband), there
443 could be some problem while insmod o2iblnd (mismatch CRC of
445 If extra Module.symvers is supported in kernel (i.e, 2.6.17),
446 this link provides solution:
447 https://bugs.openfabrics.org/show_bug.cgi?id=355
448 if extra Module.symvers is not supported in kernel, we will
449 have to run the script in bug 12316 to update
450 $LINUX/module.symvers before building o2iblnd.
451 More details about this are in bug 12316.
453 ------------------------------------------------------------------------------
455 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
456 * version 1.4.10 / 1.6.0
457 * Support for networks:
458 socklnd - kernels up to 2.6.16,
459 qswlnd - Qsnet kernel modules 5.20 and later,
460 openiblnd - IbGold 1.8.2,
462 viblnd - Voltaire ibhost 3.4.5 and later,
463 ciblnd - Topspin 3.2.0,
464 iiblnd - Infiniserv 3.3 + PathBits patch,
465 gmlnd - GM 2.1.22 and later,
466 mxlnd - MX 1.2.1 or later,
467 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
471 Description: Ptllnd didn't init kptllnd_data.kptl_idle_txs before it could be
472 possibly accessed in kptllnd_shutdown. Ptllnd should init
473 kptllnd_data.kptl_ptlid2str_lock before calling kptllnd_ptlid2str.
477 Description: gmlnd ignored some transmit errors when finalizing lnet messages.
481 Description: ptllnd logs a piece of incorrect debug info in kptllnd_peer_handle_hello.
485 Description: the_lnet.ln_finalizing was not set when the current thread is
486 about to complete messages. It only affects multi-threaded
492 Description: Changed the default kqswlnd ntxmsg=512
497 Description: Assertion failure in kernel ptllnd caused by posting passive
498 bulk buffers before connection establishment complete.
503 Description: A race in kernel ptllnd between deleting a peer and posting
504 new communications for it could hang communications -
505 manifesting as "Unexpectedly long timeout" messages.
510 Description: Kernel ptllnd lock ordering issue could hang a node.
515 Description: node crash on socket teardown race
518 Frequency : 'lctl peer_list' issued on a mx net
520 Description: Enable lctl's peer_list for MXLND
523 Frequency : after Ptllnd timeouts and portals congestion
525 Description: Credit overflows
526 Details : This was a bug in ptllnd connection establishment. The fix
527 implements better peer stamps to disambiguate connection
528 establishment and ensure both peers enter the credit flow
529 state machine consistently.
534 Description: kptllnd didn't propagate some network errors up to LNET
535 Details : This bug was spotted while investigating 11394. The fix
536 ensures network errors on sends and bulk transfers are
537 propagated to LNET/lustre correctly.
539 Severity : enhancement
541 Description: Fixed console chatter in case of -ETIMEDOUT.
543 Severity : enhancement
545 Description: Added D_NETTRACE for recording network packet history
546 (initially only for ptllnd). Also a separate userspace
547 ptllnd facility to gather history which should really be
548 covered by D_NETTRACE too, if only CDEBUG recorded history in
554 Description: o2iblnd handle early RDMA_CM_EVENT_DISCONNECTED.
555 Details : If the fabric is lossy, an RDMA_CM_EVENT_DISCONNECTED
556 callback can occur before a connection has actually been
557 established. This caused an assertion failure previously.
559 Severity : enhancement
561 Description: Multiple instances for o2iblnd
562 Details : Allow multiple instances of o2iblnd to enable networking over
563 multiple HCAs and routing between them.
567 Description: lnet deadlock in router_checker
568 Details : turned ksnd_connd_lock, ksnd_reaper_lock, and ksock_net_t:ksnd_lock
569 into BH locks to eliminate potential deadlock caused by
570 ksocknal_data_ready() preempting code holding these locks.
574 Description: Millions of failed socklnd connection attempts cause a very slow FS
575 Details : added a new route flag ksnr_scheduled to distinguish from
576 ksnr_connecting, so that a peer connection request is only turned
577 down for race concerns when an active connection to the same peer
578 is under progress (instead of just being scheduled).
580 ------------------------------------------------------------------------------
582 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
584 * Support for networks:
585 socklnd - kernels up to 2.6.16
586 qswlnd - Qsnet kernel modules 5.20 and later
587 openiblnd - IbGold 1.8.2
589 viblnd - Voltaire ibhost 3.4.5 and later
590 ciblnd - Topspin 3.2.0
591 iiblnd - Infiniserv 3.3 + PathBits patch
592 gmlnd - GM 2.1.22 and later
593 mxlnd - MX 1.2.1 or later
594 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
597 Severity : major on XT3
599 Description: libcfs overwrites /proc/sys/portals
600 Details : libcfs created a symlink from /proc/sys/portals to
601 /proc/sys/lnet for backwards compatibility. This is no
602 longer required and makes the Cray portals /proc variables
607 Description: OFED FMR API change
608 Details : This changes parameter usage to reflect a change in
609 ib_fmr_pool_map_phys() between OFED 1.0 and OFED 1.1. Note
610 that FMR support is only used in experimental versions of the
611 o2iblnd - this change does not affect standard usage at all.
613 Severity : enhancement
615 Description: new ko2iblnd module parameter: ib_mtu
616 Details : the default IB MTU of 2048 performs badly on 23108 Tavor
617 HCAs. You can avoid this problem by setting the MTU to 1024
618 using this module parameter.
620 Severity : enhancement
621 Bugzilla : 11118/11620
622 Description: ptllnd small request message buffer alignment fix
623 Details : Set the PTL_MD_LOCAL_ALIGN8 option on small message receives.
624 Round up small message size on sends in case this option
625 is not supported. 11620 was a defect in the initial
626 implementation which effectively asserted all peers had to be
627 running the correct protocol version which was fixed by always
628 NAK-ing such requests and handling any misalignments they
633 Description: When kib(nal|lnd)_del_peer() is called upon a peer whose
634 ibp_tx_queue is not empty, kib(nal|lnd)_destroy_peer()'s
635 'LASSERT(list_empty(&peer->ibp_tx_queue))' will fail.
637 Severity : enhancement
639 Description: Patchless ZC(zero copy) socklnd
640 Details : New protocol for socklnd, socklnd can support zero copy without
641 kernel patch, it's compatible with old socklnd. Checksum is
642 moved from tunables to modparams.
646 Description: When ksocknal_del_peer() is called upon a peer whose
647 ksnp_tx_queue is not empty, ksocknal_destroy_peer()'s
648 'LASSERT(list_empty(&peer->ksnp_tx_queue))' will fail.
651 Frequency : when ptlrpc is under heavy use and runs out of request buffer
653 Description: In lnet_match_blocked_msg(), md can be used without holding a
657 Frequency : very rarely
659 Description: If ksocknal_lib_setup_sock() fails, a ref on peer is lost.
660 If connd connects a route which has been closed by
661 ksocknal_shutdown(), ksocknal_create_routes() may create new
662 routes which hold references on the peer, causing shutdown
663 process to wait for peer to disappear forever.
665 Severity : enhancement
667 Description: Dump XT3 portals traces on kptllnd timeout
668 Details : Set the kptllnd module parameter "ptltrace_on_timeout=1" to
669 dump Cray portals debug traces to a file. The kptllnd module
670 parameter "ptltrace_basename", default "/tmp/lnet-ptltrace",
671 is the basename of the dump file.
674 Frequency : infrequent
676 Description: kernel ptllnd fix bug in connection re-establishment
677 Details : Kernel ptllnd could produce protocol errors e.g. illegal
678 matchbits and/or violate the credit flow protocol when trying
679 to re-establish a connection with a peer after an error or
682 Severity : enhancement
684 Description: Allow /proc/sys/lnet/debug to be set symbolically
685 Details : Allow debug and subsystem debug values to be read/set by name
686 in addition to numerically, for ease of use.
689 Frequency : only in configurations with LNET routers
691 Description: routes automatically marked down and recovered
692 Details : In configurations with LNET routers if a router fails routers
693 now actively try to recover routes that are down, unless they
694 are marked down by an administrator.
696 ------------------------------------------------------------------------------
698 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
701 Frequency : very rarely, in configurations with LNET routers and TCP
703 Description: incorrect data written to files on OSTs
704 Details : In certain high-load conditions incorrect data may be written
705 to files on the OST when using TCP networks.
707 ------------------------------------------------------------------------------
709 2006-07-31 Cluster File Systems, Inc. <info@clusterfs.com>
711 - rework CDEBUG messages rate-limiting mechanism b=10375
712 - add per-socket tunables for socklnd if the kernel is patched b=10327
714 ------------------------------------------------------------------------------
716 2006-02-15 Cluster File Systems, Inc. <info@clusterfs.com>
718 - fix use of portals/lnet pid to avoid dropping RPCs b=10074
719 - iiblnd wasn't mapping all memory, resulting in comms errors b=9776
720 - quiet LNET startup LNI message for liblustre b=10128
721 - Better console error messages if 'ip2nets' can't match an IP address
722 - Fixed overflow/use-before-set bugs in linux-time.h
723 - Fixed ptllnd bug that wasn't initialising rx descriptors completely
724 - LNET teardown failed an assertion about the route table being empty
725 - Fixed a crash in LNetEQPoll(<invalid handle>)
726 - Future protocol compatibility work (b_rls146_lnetprotovrsn)
727 - improve debug message for liblustre/Catamount nodes (b=10116)
729 2005-10-10 Cluster File Systems, Inc. <info@clusterfs.com>
730 * Configuration change for the XT3
731 The PTLLND is now used to run Lustre over Portals on the XT3.
732 The configure option(s) --with-cray-portals are no longer
733 used. Rather --with-portals=<path-to-portals-includes> is
734 used to enable building on the XT3. In addition to enable
735 XT3 specific features the option --enable-cray-xt3 must be
738 2005-10-10 Cluster File Systems, Inc. <info@clusterfs.com>
739 * Portals has been removed, replaced by LNET.
740 LNET is new networking infrastructure for Lustre, it includes a
741 reorganized network configuration mode (see the user
742 documentation for full details) as well as support for routing
743 between different network fabrics. Lustre Networking Devices
744 (LNDS) for the supported network fabrics have also been created
745 for this new infrastructure.
747 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
752 Frequency : rare (large Voltaire clusters only)
754 Description: the default number of reserved transmit descriptors was too low
755 for some large clusters
756 Details : As a workaround, the number was increased. A proper fix includes
759 2005-06-02 Cluster File Systems, Inc. <info@clusterfs.com>
764 Frequency : occasional (large-scale events, cluster reboot, network failure)
766 Description: too many error messages on console obscure actual problem and
767 can slow down/panic server, or cause recovery to fail repeatedly
768 Details : enable rate-limiting of console error messages, and some messages
769 that were console errors now only go to the kernel log
771 Severity : enhancement
773 Description: add /proc/sys/portals/catastrophe entry which will report if
774 that node has previously LBUGged
776 2005-04-06 Cluster File Systems, Inc. <info@clusterfs.com>
778 - update gmnal to use PTL_MTU, fix module refcounting (b=5786)
780 2005-04-04 Cluster File Systems, Inc. <info@clusterfs.com>
782 - handle error return code in kranal_check_fma_rx() (5915,6054)
784 2005-02-04 Cluster File Systems, Inc. <info@clusterfs.com>
786 - update vibnal (Voltaire IB NAL)
787 - update gmnal (Myrinet NAL), gmnalid
789 2005-02-04 Eric Barton <eeb@bartonsoftware.com>
791 * Landed portals:b_port_step as follows...
793 - removed CFS_DECL_SPIN*
794 just use 'spinlock_t' and initialise with spin_lock_init()
796 - removed CFS_DECL_MUTEX*
797 just use 'struct semaphore' and initialise with init_mutex()
799 - removed CFS_DECL_RWSEM*
800 just use 'struct rw_semaphore' and initialise with init_rwsem()
802 - renamed cfs_sleep_chan -> cfs_waitq
803 cfs_sleep_link -> cfs_waitlink
805 - fixed race in linux version of arch-independent socknal
806 (the ENOMEM/EAGAIN decision).
808 - Didn't fix problems in Darwin version of arch-independent socknal
809 (resetting socket callbacks, eager ack hack, ENOMEM/EAGAIN decision)
811 - removed libcfs types from non-socknal header files (only some types
812 in the header files had been changed; the .c files hadn't been