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: One down Lustre FS hangs ALL mounted Lustre filesystems
23 Details : Shared routing enhancements - peer health detection.
25 Severity : enhancement
27 Description: acceptor.c cleanup
28 Details : Code duplication in acceptor.c for the cases of kernel and
29 user-space removed. User-space libcfs tcpip primitives
30 uniformed to have prototypes similar to kernel ones. Minor
31 cosmetic changes in usocklnd to use cfs_socket_t as
32 representation of socket.
36 Description: IB path MTU mistakenly set to 1st path MTU when ib_mtu is off
37 Details : See comment 46 in bug 11245 for details - it's indeed a bug
38 introduced by the original 11245 fix.
42 Description: uptllnd credit overflow fix
43 Details : kptl_msg_t::ptlm_credits could be overflown by uptllnd since
48 Description: socklnd protocol version 3
49 Details : With current protocol V2, connections on router can be
50 blocked and can't receive any incoming messages when there is no
51 more router buffer, so ZC-ACK can't be handled (LNet message
52 can't be finalized) and will cause deadlock on router.
53 Protocol V3 has a dedicated connection for emergency messages
54 like ZC-ACK to router, messages on this dedicated connection
55 don't need any credit so will never be blocked. Also, V3 can send
56 keepalive ping in specified period for router healthy checking.
58 -------------------------------------------------------------------------------
60 12-31-2008 Sun Microsystems, Inc.
62 * Support for networks:
63 socklnd - any kernel supported by Lustre,
64 qswlnd - Qsnet kernel modules 5.20 and later,
65 openiblnd - IbGold 1.8.2,
66 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
67 viblnd - Voltaire ibhost 3.4.5 and later,
68 ciblnd - Topspin 3.2.0,
69 iiblnd - Infiniserv 3.3 + PathBits patch,
70 gmlnd - GM 2.1.22 and later,
71 mxlnd - MX 1.2.1 or later,
72 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
76 Description: workaround for OOM from o2iblnd
77 Details : OFED needs allocate big chunk of memory for QP while creating
78 connection for o2iblnd, OOM can happen if no such a contiguous
80 QP size is decided by concurrent_sends and max_fragments of
81 o2iblnd, now we permit user to specify smaller value for
82 concurrent_sends of o2iblnd(i.e: concurrent_sends=7), which
83 will decrease memory block size required by creating QP.
87 Description: Support Zerocopy receive of Chelsio device
88 Details : Chelsio driver can support zerocopy for iov[1] if it's
89 contiguous and large enough.
93 Description: fix credit flow deadlock in uptllnd
97 Description: finalize network operation in reasonable time
98 Details : conf-sanity test_32a couldn't stop ost and mds because it
99 tried to access non-existent peer and tcp connect took
100 quite long before timing out.
104 Description: Continuous recovery on 33 of 413 nodes after lustre oss failure
105 Details : Lost reference on conn prevents peer from being destroyed, which
106 could prevent new peer creation if peer count has reached upper
111 Description: LNET Selftest results in Soft lockup on OSS CPU
112 Details : only hits when 8 or more o2ib clients involved and a session is
113 torn down with 'lst end_session' without preceeding 'lst stop'.
117 Description: concurrent_sends in IB LNDs should not be changeable at run time
118 Details : concurrent_sends in IB LNDs should not be changeable at run time
122 Description: ptl_send_rpc hits LASSERT when ptl_send_buf fails
123 Details : only hits under out-of-memory situations
126 -------------------------------------------------------------------------------
128 2009-02-07 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
143 Description: workaround for OOM from o2iblnd
144 Details : OFED needs allocate big chunk of memory for QP while creating
145 connection for o2iblnd, OOM can happen if no such a contiguous
147 QP size is decided by concurrent_sends and max_fragments of
148 o2iblnd, now we permit user to specify smaller value for
149 concurrent_sends of o2iblnd(i.e: concurrent_sends=7), which
150 will decrease memory block size required by creating QP.
154 Description: Support Zerocopy receive of Chelsio device
155 Details : Chelsio driver can support zerocopy for iov[1] if it's
156 contiguous and large enough.
159 Description: fix credit flow deadlock in uptllnd
163 Description: finalize network operation in reasonable time
164 Details : conf-sanity test_32a couldn't stop ost and mds because it
165 tried to access non-existent peer and tcp connect took
166 quite long before timing out.
170 Description: Continuous recovery on 33 of 413 nodes after lustre oss failure
171 Details : Lost reference on conn prevents peer from being destroyed, which
172 could prevent new peer creation if peer count has reached upper
177 Description: LNET Selftest results in Soft lockup on OSS CPU
178 Details : only hits when 8 or more o2ib clients involved and a session is
179 torn down with 'lst end_session' without preceeding 'lst stop'.
183 Description: concurrent_sends in IB LNDs should not be changeable at run time
184 Details : concurrent_sends in IB LNDs should not be changeable at run time
186 -------------------------------------------------------------------------------
188 11-03-2008 Sun Microsystems, Inc.
190 * Support for networks:
191 socklnd - any kernel supported by Lustre,
192 qswlnd - Qsnet kernel modules 5.20 and later,
193 openiblnd - IbGold 1.8.2,
194 o2iblnd - OFED 1.1, 1.2.0, 1.2.5, and 1.3
195 viblnd - Voltaire ibhost 3.4.5 and later,
196 ciblnd - Topspin 3.2.0,
197 iiblnd - Infiniserv 3.3 + PathBits patch,
198 gmlnd - GM 2.1.22 and later,
199 mxlnd - MX 1.2.1 or later,
200 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
204 Description: ptl_send_rpc hits LASSERT when ptl_send_buf fails
205 Details : only hits under out-of-memory situations
208 -------------------------------------------------------------------------------
211 04-26-2008 Sun Microsystems, Inc.
213 * Support for networks:
214 socklnd - any kernel supported by Lustre,
215 qswlnd - Qsnet kernel modules 5.20 and later,
216 openiblnd - IbGold 1.8.2,
217 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5
218 viblnd - Voltaire ibhost 3.4.5 and later,
219 ciblnd - Topspin 3.2.0,
220 iiblnd - Infiniserv 3.3 + PathBits patch,
221 gmlnd - GM 2.1.22 and later,
222 mxlnd - MX 1.2.1 or later,
223 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
227 Description: excessive debug information removed
228 Details : excessive debug information removed
232 Description: ksocknal_create_conn() hit ASSERTION during connection race
233 Details : ksocknal_create_conn() hit ASSERTION during connection race
237 Description: ksocknal_send_hello() hit ASSERTION while connecting race
238 Details : ksocknal_send_hello() hit ASSERTION while connecting race
242 Description: o2iblnd/ptllnd credit deadlock in a routed config.
243 Details : o2iblnd/ptllnd credit deadlock in a routed config.
247 Description: High load after starting lnet
248 Details : gmlnd should sleep in rx thread in interruptible way. Otherwise,
249 uptime utility reports high load that looks confusingly.
253 Description: ksocklnd fails to establish connection if accept_port is high
254 Details : PID remapping must not be done for active (outgoing) connections
256 --------------------------------------------------------------------------------
258 2008-01-11 Sun Microsystems, Inc.
260 * Support for networks:
261 socklnd - any kernel supported by Lustre,
262 qswlnd - Qsnet kernel modules 5.20 and later,
263 openiblnd - IbGold 1.8.2,
264 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5
265 viblnd - Voltaire ibhost 3.4.5 and later,
266 ciblnd - Topspin 3.2.0,
267 iiblnd - Infiniserv 3.3 + PathBits patch,
268 gmlnd - GM 2.1.22 and later,
269 mxlnd - MX 1.2.1 or later,
270 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
273 Description: liblustre network error
274 Details : liblustre clients should understand LNET_ACCEPT_PORT environment
275 variable even if they don't start lnet acceptor.
279 Description: Strange message from lnet (Ignoring prediction from the future)
280 Details : Incorrect calculation of peer's last_alive value in ksocklnd
282 --------------------------------------------------------------------------------
284 2007-12-07 Cluster File Systems, Inc. <info@clusterfs.com>
286 * Support for networks:
287 socklnd - any kernel supported by Lustre,
288 qswlnd - Qsnet kernel modules 5.20 and later,
289 openiblnd - IbGold 1.8.2,
290 o2iblnd - OFED 1.1 and 1.2.0, 1.2.5.
291 viblnd - Voltaire ibhost 3.4.5 and later,
292 ciblnd - Topspin 3.2.0,
293 iiblnd - Infiniserv 3.3 + PathBits patch,
294 gmlnd - GM 2.1.22 and later,
295 mxlnd - MX 1.2.1 or later,
296 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
300 Description: ASSERTION(me == md->md_me) failed in lnet_match_md()
304 Description: increase send queue size for ciblnd/openiblnd
308 Description: new userspace socklnd
309 Details : Old userspace tcpnal that resided in lnet/ulnds/socklnd replaced
310 with new one - usocklnd.
312 Severity : enhancement
314 Description: Console message flood
315 Details : Make cdls ratelimiting more tunable by adding several tunable in
316 procfs /proc/sys/lnet/console_{min,max}_delay_centisecs and
317 /proc/sys/lnet/console_backoff.
319 --------------------------------------------------------------------------------
321 2007-09-27 Cluster File Systems, Inc. <info@clusterfs.com>
323 * Support for networks:
324 socklnd - any kernel supported by Lustre,
325 qswlnd - Qsnet kernel modules 5.20 and later,
326 openiblnd - IbGold 1.8.2,
327 o2iblnd - OFED 1.1 and 1.2,
328 viblnd - Voltaire ibhost 3.4.5 and later,
329 ciblnd - Topspin 3.2.0,
330 iiblnd - Infiniserv 3.3 + PathBits patch,
331 gmlnd - GM 2.1.22 and later,
332 mxlnd - MX 1.2.1 or later,
333 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
337 Description: /proc/sys/lnet has non-sysctl entries
338 Details : Updating dump_kernel/daemon_file/debug_mb to use sysctl variables
342 Description: TOE Kernel panic by ksocklnd
343 Details : offloaded sockets provide their own implementation of sendpage,
344 can't call tcp_sendpage() directly
348 Description: kibnal_shutdown() doesn't finish; lconf --cleanup hangs
349 Details : races between lnd_shutdown and peer creation prevent
350 lnd_shutdown from finishing.
354 Description: open files rlimit 1024 reached while liblustre testing
355 Details : ulnds/socklnd must close open socket after unsuccessful
360 Description: build error
361 Details : fix typos in gmlnd, ptllnd and viblnd
363 ------------------------------------------------------------------------------
365 2007-07-30 Cluster File Systems, Inc. <info@clusterfs.com>
367 * Support for networks:
368 socklnd - kernels up to 2.6.16,
369 qswlnd - Qsnet kernel modules 5.20 and later,
370 openiblnd - IbGold 1.8.2,
371 o2iblnd - OFED 1.1 and 1.2
372 viblnd - Voltaire ibhost 3.4.5 and later,
373 ciblnd - Topspin 3.2.0,
374 iiblnd - Infiniserv 3.3 + PathBits patch,
375 gmlnd - GM 2.1.22 and later,
376 mxlnd - MX 1.2.1 or later,
377 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
379 2007-06-21 Cluster File Systems, Inc. <info@clusterfs.com>
381 * Support for networks:
382 socklnd - kernels up to 2.6.16,
383 qswlnd - Qsnet kernel modules 5.20 and later,
384 openiblnd - IbGold 1.8.2,
386 viblnd - Voltaire ibhost 3.4.5 and later,
387 ciblnd - Topspin 3.2.0,
388 iiblnd - Infiniserv 3.3 + PathBits patch,
389 gmlnd - GM 2.1.22 and later,
390 mxlnd - MX 1.2.1 or later,
391 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
395 Description: Initialize cpumask before use
399 Description: ASSERTION failures when upgrading to the patchless zero-copy
401 Details : This bug affects "rolling upgrades", causing an inconsistent
402 protocol version negotiation and subsequent assertion failure
403 during rolling upgrades after the first wave of upgrades.
407 Details : Change "dropped message" CERRORs to D_NETERROR so they are
408 logged instead of creating "console chatter" when a lustre
409 timeout races with normal RPC completion.
412 Details : lnet_clear_peer_table can wait forever if user forgets to
416 Details : libcfs_id2str should check pid against LNET_PID_ANY.
420 Description: added LNET self test
421 Details : landing b_self_test
426 Description: cfs_duration_{u,n}sec() wrongly calculate nanosecond part of
428 Details : do_div() macro is used incorrectly.
430 2007-04-23 Cluster File Systems, Inc. <info@clusterfs.com>
434 Description: make panic on lbug configurable
438 Description: Add OFED1.2 support to o2iblnd
439 Details : o2iblnd depends on OFED's modules, if out-tree OFED's modules
440 are installed (other than kernel's in-tree infiniband), there
441 could be some problem while insmod o2iblnd (mismatch CRC of
443 If extra Module.symvers is supported in kernel (i.e, 2.6.17),
444 this link provides solution:
445 https://bugs.openfabrics.org/show_bug.cgi?id=355
446 if extra Module.symvers is not supported in kernel, we will
447 have to run the script in bug 12316 to update
448 $LINUX/module.symvers before building o2iblnd.
449 More details about this are in bug 12316.
451 ------------------------------------------------------------------------------
453 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
454 * version 1.4.10 / 1.6.0
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: Ptllnd didn't init kptllnd_data.kptl_idle_txs before it could be
470 possibly accessed in kptllnd_shutdown. Ptllnd should init
471 kptllnd_data.kptl_ptlid2str_lock before calling kptllnd_ptlid2str.
475 Description: gmlnd ignored some transmit errors when finalizing lnet messages.
479 Description: ptllnd logs a piece of incorrect debug info in kptllnd_peer_handle_hello.
483 Description: the_lnet.ln_finalizing was not set when the current thread is
484 about to complete messages. It only affects multi-threaded
490 Description: Changed the default kqswlnd ntxmsg=512
495 Description: Assertion failure in kernel ptllnd caused by posting passive
496 bulk buffers before connection establishment complete.
501 Description: A race in kernel ptllnd between deleting a peer and posting
502 new communications for it could hang communications -
503 manifesting as "Unexpectedly long timeout" messages.
508 Description: Kernel ptllnd lock ordering issue could hang a node.
513 Description: node crash on socket teardown race
516 Frequency : 'lctl peer_list' issued on a mx net
518 Description: Enable lctl's peer_list for MXLND
521 Frequency : after Ptllnd timeouts and portals congestion
523 Description: Credit overflows
524 Details : This was a bug in ptllnd connection establishment. The fix
525 implements better peer stamps to disambiguate connection
526 establishment and ensure both peers enter the credit flow
527 state machine consistently.
532 Description: kptllnd didn't propagate some network errors up to LNET
533 Details : This bug was spotted while investigating 11394. The fix
534 ensures network errors on sends and bulk transfers are
535 propagated to LNET/lustre correctly.
537 Severity : enhancement
539 Description: Fixed console chatter in case of -ETIMEDOUT.
541 Severity : enhancement
543 Description: Added D_NETTRACE for recording network packet history
544 (initially only for ptllnd). Also a separate userspace
545 ptllnd facility to gather history which should really be
546 covered by D_NETTRACE too, if only CDEBUG recorded history in
552 Description: o2iblnd handle early RDMA_CM_EVENT_DISCONNECTED.
553 Details : If the fabric is lossy, an RDMA_CM_EVENT_DISCONNECTED
554 callback can occur before a connection has actually been
555 established. This caused an assertion failure previously.
557 Severity : enhancement
559 Description: Multiple instances for o2iblnd
560 Details : Allow multiple instances of o2iblnd to enable networking over
561 multiple HCAs and routing between them.
565 Description: lnet deadlock in router_checker
566 Details : turned ksnd_connd_lock, ksnd_reaper_lock, and ksock_net_t:ksnd_lock
567 into BH locks to eliminate potential deadlock caused by
568 ksocknal_data_ready() preempting code holding these locks.
572 Description: Millions of failed socklnd connection attempts cause a very slow FS
573 Details : added a new route flag ksnr_scheduled to distinguish from
574 ksnr_connecting, so that a peer connection request is only turned
575 down for race concerns when an active connection to the same peer
576 is under progress (instead of just being scheduled).
578 ------------------------------------------------------------------------------
580 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
582 * Support for networks:
583 socklnd - kernels up to 2.6.16
584 qswlnd - Qsnet kernel modules 5.20 and later
585 openiblnd - IbGold 1.8.2
587 viblnd - Voltaire ibhost 3.4.5 and later
588 ciblnd - Topspin 3.2.0
589 iiblnd - Infiniserv 3.3 + PathBits patch
590 gmlnd - GM 2.1.22 and later
591 mxlnd - MX 1.2.1 or later
592 ptllnd - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
595 Severity : major on XT3
597 Description: libcfs overwrites /proc/sys/portals
598 Details : libcfs created a symlink from /proc/sys/portals to
599 /proc/sys/lnet for backwards compatibility. This is no
600 longer required and makes the Cray portals /proc variables
605 Description: OFED FMR API change
606 Details : This changes parameter usage to reflect a change in
607 ib_fmr_pool_map_phys() between OFED 1.0 and OFED 1.1. Note
608 that FMR support is only used in experimental versions of the
609 o2iblnd - this change does not affect standard usage at all.
611 Severity : enhancement
613 Description: new ko2iblnd module parameter: ib_mtu
614 Details : the default IB MTU of 2048 performs badly on 23108 Tavor
615 HCAs. You can avoid this problem by setting the MTU to 1024
616 using this module parameter.
618 Severity : enhancement
619 Bugzilla : 11118/11620
620 Description: ptllnd small request message buffer alignment fix
621 Details : Set the PTL_MD_LOCAL_ALIGN8 option on small message receives.
622 Round up small message size on sends in case this option
623 is not supported. 11620 was a defect in the initial
624 implementation which effectively asserted all peers had to be
625 running the correct protocol version which was fixed by always
626 NAK-ing such requests and handling any misalignments they
631 Description: When kib(nal|lnd)_del_peer() is called upon a peer whose
632 ibp_tx_queue is not empty, kib(nal|lnd)_destroy_peer()'s
633 'LASSERT(list_empty(&peer->ibp_tx_queue))' will fail.
635 Severity : enhancement
637 Description: Patchless ZC(zero copy) socklnd
638 Details : New protocol for socklnd, socklnd can support zero copy without
639 kernel patch, it's compatible with old socklnd. Checksum is
640 moved from tunables to modparams.
644 Description: When ksocknal_del_peer() is called upon a peer whose
645 ksnp_tx_queue is not empty, ksocknal_destroy_peer()'s
646 'LASSERT(list_empty(&peer->ksnp_tx_queue))' will fail.
649 Frequency : when ptlrpc is under heavy use and runs out of request buffer
651 Description: In lnet_match_blocked_msg(), md can be used without holding a
655 Frequency : very rarely
657 Description: If ksocknal_lib_setup_sock() fails, a ref on peer is lost.
658 If connd connects a route which has been closed by
659 ksocknal_shutdown(), ksocknal_create_routes() may create new
660 routes which hold references on the peer, causing shutdown
661 process to wait for peer to disappear forever.
663 Severity : enhancement
665 Description: Dump XT3 portals traces on kptllnd timeout
666 Details : Set the kptllnd module parameter "ptltrace_on_timeout=1" to
667 dump Cray portals debug traces to a file. The kptllnd module
668 parameter "ptltrace_basename", default "/tmp/lnet-ptltrace",
669 is the basename of the dump file.
672 Frequency : infrequent
674 Description: kernel ptllnd fix bug in connection re-establishment
675 Details : Kernel ptllnd could produce protocol errors e.g. illegal
676 matchbits and/or violate the credit flow protocol when trying
677 to re-establish a connection with a peer after an error or
680 Severity : enhancement
682 Description: Allow /proc/sys/lnet/debug to be set symbolically
683 Details : Allow debug and subsystem debug values to be read/set by name
684 in addition to numerically, for ease of use.
687 Frequency : only in configurations with LNET routers
689 Description: routes automatically marked down and recovered
690 Details : In configurations with LNET routers if a router fails routers
691 now actively try to recover routes that are down, unless they
692 are marked down by an administrator.
694 ------------------------------------------------------------------------------
696 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
699 Frequency : very rarely, in configurations with LNET routers and TCP
701 Description: incorrect data written to files on OSTs
702 Details : In certain high-load conditions incorrect data may be written
703 to files on the OST when using TCP networks.
705 ------------------------------------------------------------------------------
707 2006-07-31 Cluster File Systems, Inc. <info@clusterfs.com>
709 - rework CDEBUG messages rate-limiting mechanism b=10375
710 - add per-socket tunables for socklnd if the kernel is patched b=10327
712 ------------------------------------------------------------------------------
714 2006-02-15 Cluster File Systems, Inc. <info@clusterfs.com>
716 - fix use of portals/lnet pid to avoid dropping RPCs b=10074
717 - iiblnd wasn't mapping all memory, resulting in comms errors b=9776
718 - quiet LNET startup LNI message for liblustre b=10128
719 - Better console error messages if 'ip2nets' can't match an IP address
720 - Fixed overflow/use-before-set bugs in linux-time.h
721 - Fixed ptllnd bug that wasn't initialising rx descriptors completely
722 - LNET teardown failed an assertion about the route table being empty
723 - Fixed a crash in LNetEQPoll(<invalid handle>)
724 - Future protocol compatibility work (b_rls146_lnetprotovrsn)
725 - improve debug message for liblustre/Catamount nodes (b=10116)
727 2005-10-10 Cluster File Systems, Inc. <info@clusterfs.com>
728 * Configuration change for the XT3
729 The PTLLND is now used to run Lustre over Portals on the XT3.
730 The configure option(s) --with-cray-portals are no longer
731 used. Rather --with-portals=<path-to-portals-includes> is
732 used to enable building on the XT3. In addition to enable
733 XT3 specific features the option --enable-cray-xt3 must be
736 2005-10-10 Cluster File Systems, Inc. <info@clusterfs.com>
737 * Portals has been removed, replaced by LNET.
738 LNET is new networking infrastructure for Lustre, it includes a
739 reorganized network configuration mode (see the user
740 documentation for full details) as well as support for routing
741 between different network fabrics. Lustre Networking Devices
742 (LNDS) for the supported network fabrics have also been created
743 for this new infrastructure.
745 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
750 Frequency : rare (large Voltaire clusters only)
752 Description: the default number of reserved transmit descriptors was too low
753 for some large clusters
754 Details : As a workaround, the number was increased. A proper fix includes
757 2005-06-02 Cluster File Systems, Inc. <info@clusterfs.com>
762 Frequency : occasional (large-scale events, cluster reboot, network failure)
764 Description: too many error messages on console obscure actual problem and
765 can slow down/panic server, or cause recovery to fail repeatedly
766 Details : enable rate-limiting of console error messages, and some messages
767 that were console errors now only go to the kernel log
769 Severity : enhancement
771 Description: add /proc/sys/portals/catastrophe entry which will report if
772 that node has previously LBUGged
774 2005-04-06 Cluster File Systems, Inc. <info@clusterfs.com>
776 - update gmnal to use PTL_MTU, fix module refcounting (b=5786)
778 2005-04-04 Cluster File Systems, Inc. <info@clusterfs.com>
780 - handle error return code in kranal_check_fma_rx() (5915,6054)
782 2005-02-04 Cluster File Systems, Inc. <info@clusterfs.com>
784 - update vibnal (Voltaire IB NAL)
785 - update gmnal (Myrinet NAL), gmnalid
787 2005-02-04 Eric Barton <eeb@bartonsoftware.com>
789 * Landed portals:b_port_step as follows...
791 - removed CFS_DECL_SPIN*
792 just use 'spinlock_t' and initialise with spin_lock_init()
794 - removed CFS_DECL_MUTEX*
795 just use 'struct semaphore' and initialise with init_mutex()
797 - removed CFS_DECL_RWSEM*
798 just use 'struct rw_semaphore' and initialise with init_rwsem()
800 - renamed cfs_sleep_chan -> cfs_waitq
801 cfs_sleep_link -> cfs_waitlink
803 - fixed race in linux version of arch-independent socknal
804 (the ENOMEM/EAGAIN decision).
806 - Didn't fix problems in Darwin version of arch-independent socknal
807 (resetting socket callbacks, eager ack hack, ENOMEM/EAGAIN decision)
809 - removed libcfs types from non-socknal header files (only some types
810 in the header files had been changed; the .c files hadn't been