*************************************************** ** NP00640 FILE ** ** DATE = 08/08/96 ** PHONE = 8-444-0017 ** CONTACT = SANDY PETERSON ** C/T = RD0011 ** USER = SANDYPE NODE = RALVM29 ** *************************************************** APARTXT NA02340 = test mask data base being accessed when test frame filters not enabled - causing unnecessary work to be done APARTXT NA01748 = netbios broadcasts are filling the analyst queue within dls resulting in an out of resource condition. Frames are getting picked off a queue by the analyst every 14 seconds, resulting in timed out conditions resulting in additional broadcasts, making the condition worse. APARTXT NA01792 = Build and install changes APARTXT NA02433 = arp table size not correct STOP_SYMPTOM APARTXT NA01997 = Internal Enhancements for NP00601 APARTXT NA02213 = Add -M option to ifstat for mib variable display STOP_SYMPTOM APARTXT NA02219 = The maximum number of conversations that DLSw can have active at any one time is 5000. With the current design of DLSw, this is an unreasonable number of conversations that can be supported. The number of conversations should be reduced to 1000. This is more reasonable, plus will reduce the size of the Conversation Table from 350,000 bytes to 70,000 bytes, saving valuable memory space. More space will also be saved in the reduction of the CONVINDEX database, plus add some efficiency to some of the DLSw routines. APARTXT NA02003 = Internal enhancements APARTXT NA02011 = Wandeamon teminates when a except takes to long. APARTXT NA02281 = If a start_ls fails due to circuit already established, HALT frame might go to wrong partner APARTXT NA02220 = If two routers on the same Frame Relay link are configured for the same AppleTalk node address, the probe logic fails to resolve this address conflict. You end up with two routers set to something like 500.1 instead of 500.1 on one and 500.2 on the second guy up. APARTXT NA02434 = causes bridged IP to intermittently HANG Independant arp packets were being dropped as dups when translation bridging due to the way the ARB Copy database prevents looping frames from being translated from SR to TB. This can cause problems with routers adjacent to this bridge. If the ARP entries time out about the same time, subsuquent arps will be prevented from going thru this bridge. A gap in ~ 2 seconds will be seen between subs arps which might cause a long delay for the last entry that timed out. APARTXT NA02285 = See above. APARTXT NA02287 = The way landaemon and slandaemon acquire shared memory for the Parnter LAN and SLAN Tables is incorrect. They accually get enough shared memory for only 1 entry in the tables. Fortunately, the minimum size of shared memory is one page, so it ends up working by mistake because the table (for now) will fit in one page. But the shmget should still be fixed to work correctly. Also, if the parms LAN_PART or SLAN_PART is removed from the /usr/lpp/dls/parms file, then the shmget of the tables will fail because the defines for D_LAN_PART/D_SLAN_PART in the dls_lan.h/dls_slan.h files are strings and not numeric values. APARTXT NA02214 = ---- NCP ---- 6611 No data gets to NCP from 6611 on T/R (non-bridged) running HPR APARTXT NA02288 = Add an option to DLSTOOLS to display the internal DLSw LAN Table and SLAN Table contents. These could be handy for debug purposes. APARTXT NA02301 = see above. APARTXT NA02031 = Customer has dual token ring. One port stops transmitting, but is still receiving. Adapter is NOT hung and is NOT marked down (in system connections). The XmitMaxTlistDrop MIB counter is increasing. APARTXT NA02182 = The following fixes were made to netstat/ifstat: 1. netstat -nr and netstat -r - Display netmask for route - Allow subnet zero routes to be displayed correctly. - Add Reject/Blackhole route flags 2. netstat -ni and netstat -i - Allow subnet zero networks to be displayed correctly - Make network field larger to allow display - Line up counters under heading correctly 3. ifstat -nr and ifstat -r - Display netmask for route - Allow subnet 0 routes to be displayed correctly 4. ifstat -ni and netstat -i - Display netmask for interface - Allow subnet zero networks to be displayed correctly - Show P2P destination for P2P links APARTXT NA02053 = X.25 takes 10 minutes to start link recovery after link goes down. APARTXT NA02183 = "recv: connection timeout" message appears on interface statistics screen when x25 interfaces are present APARTXT NA02121 = Doing translational (SRTB) bridging from token ring to frame relay DLCI that is configured as translational bridge. There are other DLCIs that are configured for source route bridge. In this case, specifically routed frames destined for the TB DLCI are actually sent as source routed frames. Sessions can not be established. APARTXT NA02221 = When there are many exports for BGP with the same from AS, there can be problems if they are specified in the wrong order. STOP_SYMPTOM APARTXT NA02184 = MIB display (ifstat -M) code was removed. ifstat -M won't work STOP_SYMPTOM APARTXT NA02166 = Add option to dlstools to display the netbios names in hex. APARTXT NA02160 = lanwrite message queue full lanwrite process blocked station transmit queue full DLSw buffer pool high utilization degraded DLSw performance APARTXT NA02189 = after pathswitch, packets sent at low priority instead of medium priority APARTXT NA02191 = Device driver check for packet size > 2200 incorrect. APARTXT NA02186 = gatedcfgd needs to generate more error info APARTXT NA02435 = snanau mib bugs APARTXT NA02187 = APPN sessions are taken down when the main memory paritition is under stress, but the buffer partition is not under stress. The new design is supposed to avoid taking down sessions due to memory stress, until the buffer memory is at least under some stress. APARTXT NA02215 = The analyst was always caching the source netbios name regardless of what the netbios command was (some netbios commands do not have valid names). This would fill up the netbios reachability database with lots of netbios name number_1 values, which are useless. Looking at the netbios command before caching the names will prevent this from occurring. APARTXT NA02170 = START_TEST ethernet dls regression using DIX stations needs to be performed to make sure this defect works. STOP_TEST APARTXT NA02190 = See above APARTXT NA02222 = not applicable, these are enhancements to a back-door debugging program. APARTXT NA02188 = f023 abend in Pwait APARTXT NA02173 = dlstools w shows partners in exiting for both WW & WR. Netstat -n |grep 206 shows only 2067 to 2065 socket for this partner. PD_Sysmon.trc shows ip connectivity loss to this partn er going up/down. Defect 19786 introduced this problem to ptf 385 per Dave B. See Dave B. Wandaemon exiting Doc. APARTXT NA02175 = smux.pppd dump when ibmpppLinkControlTable queries are more than 16 ports. This causes system utilization to go to 100% Each time an snmp ppp query is done, smux.pppd dumps. APARTXT NA02292 = poor messages and incomplete error recovery from errors on writex() and ioctl(). APARTXT NA02223 = The following fixes ensure that the lsdb_border pointer is cleared for SUM_NETs, SUM_ASBs, and ASEs when the ABR/ASBR is not "active". Otherwise, inactive ABR/ASBRs can be freed and re-allocated to other ABR/ASBRs, which can easily cause routing loops. In addition, since the lsdb_route and lsdb_ab_rtr pointers are unioned, an lsdb with a route may be confused for an active ABR and cause a core. -- Richard Woundy APARTXT NA02302 = When processing a received CapEx Msg, better error messaging is needed when the CapEx Msg received is invalid. This will become more important as the 6611 starts interoperating with other vendors' inferior DLSw function. APARTXT NA02293 = This defect will track enhancements made to dlstools, most notible: improved outputting of DLSw semaphores and message queues, displaying of DLSw buffer pool info (without going thru entire buffer pool), and output enhancements. APARTXT NA02176 = can go into a self initiated re-boot when bridging traffic that contains NetBios and SNA traffic through translatio nal bridging from T.R. to E.net if a slow(9600 in test case) serial link is involved. The following was the test/reproducibl e configuration. __ ____ ____ _____ Servers / |6611| ____ |6611| |6611| SNA | |--|TL |--| |--|TP |---//--|SR |--- NetBios _/ |Bri | |Bri | |Bri | |____| |DLS | |DLS | / |PC | |____| |____|--| |--|Cli| / |ent| |___| In the test case, the PC client tries to copy a file on the NetBios server(in the test case, a PS2) and the file size is 80K. In most cases the 6611 doing the translational bridging will fail and re-boot. Short messages do not appear to cause the problem and if the serial link is replaced with a high speed( 500K) link, the problem did not occur. The problem was al so not inducible by rapid page browsing in and SNA session. APARTXT NA02178 = CfgDaemon is attempting to refresh invalid component sap. This is causing SYSMON to reject the command and if it is recieved at the same time that CfgDaemon tries to start the interface scripts, SYSMON loses those requests because they are being read in with the invalid component. This is a problem with the IPX script handler for ipx sap filters. This code is no longer used. APARTXT NA02436 = last objects.h for r4ap, maybe APARTXT NA02192 = Inventory changes for r4ap_d36fx level ptf620 APARTXT NA02202 = looppingtest will not be called to send echos and ultimately open NCPs unless LQM is enabled. STOP_SYMPTOM APARTXT NA02224 = fastpath command tftp does not work. APARTXT NA02205 = NN's cached LUs are deleted when it processes a LOCATE APARTXT NA02216 = need dlur traps with associated dlur alerts APARTXT NA02211 = Getting SDL8053 errors and the sdlc port is not coming up. Start ed occuring when migrated from v1r2 to NP00601 using NP99917. APARTXT NA02295 = The appletalk daemon atd was growing due to a couple of memory leaks. Found at Cornell. APARTXT NA02296 = Translational bridge database entries do not learn RI's proplery for frames commong from a bulternational. The RI asved will have a bridge number of 0 prior to or subsiquent to the Designated ring number. (e.g. 80901110DED5ABC is a possible RI learned which should look like 80901115DED5ABC. DED is the designated ring. From my tests, I have observed that receiving these frames on a BNE port does not influence how it is passed to the destination ring. APARTXT NA02217 = f023 abend in xxxims00 APARTXT NA02297 = Two changes in the appletalk microcode: 1) at_tree.c: make a modification to disable interrupts when inserting or deleting nodes from the route tree. At Cornell they were experiencing ethernet cards going dead on them, until this was fixed. 2) implement Robert Jakes advice and use qamb_lifo call instead of qa_lifo. Apparently a hardware error can occur that causes too long a frame to come in that becomes a multi-buffer frame. Because AppleTalk only deals in fairly short frames, the code had not anticipated having to release multiple buffers. It can cause the adapter to crash. APARTXT NA02240 = CP-CP sessions fail 0805 session limit exceeded on redrive. APARTXT NA02234 = Trace all APPN system traces doesn't work from sysman. APARTXT NA02341 = lsifnet tool not included in r4ap. APARTXT NA02437 = t960/te-596/e596_rx.s fails to compiles # 21274 start cause failure changed to # start 21274 STOP_SYMPTOM APARTXT NA02225 = After Customer had WAN outage, some DLS SNA sessions were hung in RemDscPnd state, State 10 . This state only occurred on one side of the DLS partners. When workstations tried to login even after WAN restored, logins were rejected because RemDscPnd had not cleared for these workstations' MAC addresses. APARTXT NA02246 = Appletalk not starting on new version of release from tape because resetappltalk script depends on cfg vars that are not available when it attempts to run. This was one of the iml improvement efforts that didn't work. APARTXT NA02228 = Rinstall gives a error message "shift: bad number" if there is a blank line in the control file. This apar referes to rinstall at the 383 level. Below is a list of problems with making rinstall work at this level. 1) customer must define a host name to ip address for both the server and the client(s). On the 6611 this can be done on the config tool, node management, remote hosts 2) there is a apar na01369 against rinstall, fixed in 385, the main fix from this apar can be put onto a 383 machine APARTXT NA02298 = On r4ap, v1r4 and PTF385, when the command "interface sdlc mpq0 disable" is issued, the port is only momentarily disabled. This was observed while running SDLC traffic through a 6611 and noting that the traffic, which was apparent on the workstation screen, quickly resumed after issuing the disable command. This can also be observed by monitoring the output of "dlstools c". Ports disappear from "connected" mode after an "interface disable" -- they should not become "connected" again until an "interface sdlc mpq0 enable" (of course 0 could be any legal port number) is issued. APARTXT NA02230 = stopped communicating, and the error log is filled with system running out of paging space entries. APARTXT NA02342 = Option set 1116 APARTXT NA02299 = message: "haltimg circuit" in pd_sysmon.trc identifying an APPN circuit. Accompanying this will be a message indicating that an ioctl(DLC_EXIT_LBUSY) failed. STOP_SYMPTOM APARTXT NA02438 = Problem is abend or hang conditions, do to unexpected/invalid message processing sequences, servicing process queues. APARTXT NA02343 = Changing primary address on a port that has secondary addresses defined, does not cause the box to reboot. APARTXT NA02303 = Make sure time stamps now exist in pd_x25d.out and that nothing else breaks. APARTXT NA02245 = If you configure a router to only accept configurations from a certain host, and then send a configuration from that host, the router rejects the config. APARTXT NA02347 = aping function doesn't work on the 6611 APARTXT NA02253 = GATED taking up CPU cycles in a multiple area config. 6611 with OSPF/PPP over F/R to a 2210 using point-to-multipoint. Problem: When more than one router is advertising a stub net and the 6611 is acting as an Area Border Router, the 6611 will needlesly regenerate a summary for the stub net advertised by more than one router whenever an SPF calculation is performed. While the correct route is installed, network bandwidth and router cycles are consumed. *Customer Situation* With 2210 and point-to-multipoint over frame relay, the 2210 always advertises it's local interface with a cost of zero. Conversely, 6611 implements logical point-to-point over frame relay and always advertises the destination networks to which it is attached. So, in a topology where you have a 6611 using logical point to point over F/R to a 2210 using point to multipoint, the problem is multiplied. APARTXT NA02254 = Missing summary routes in UPS network where 3 areas are being used. Only noticed on one of the two routers using a 3rd area. The other 3 area router is a backup to this one and would not have produced the problem unless it became primary. Problem is intermittent, once a week but more often recently... Smact restart gated will restore operation for another indeterminant cycle. APARTXT NA02255 = Sessions fail to come up with 0812000D sense code in error log. Insufficient Storage. APARTXT NA02346 = f023 abend in ss APARTXT NA02256 = On sending duplicate config to router with IPX configured the config daemon is restarting the interface scripts, even though no config objects have changed. This is caused by the external variable chPT_IPX_01 having been loaded on the initial config and being concatinated with itself on the second config in handle_ipx_xns.c. This caused memory to be overwritten, corrupting another variable. APARTXT NA02257 = Internal Enhancements for NP00620 APARTXT NA02348 = route setup reply from 6611 destination does not contain ps timer APARTXT NA02258 = IP filter currently checks the dst/src ports for a given traffic which handicap its function for some requirements. Customer would like to enable a traffic (such as telnet/ftp - both requests and responses) come in one direction and block the same traffic initiated from different direction. APARTXT NA02317 = Fix framerelay dlci view for multiple IP addrs per DLCI STOP_SYMPTOM START_TEST Issue fr dlci v int for framerelay interface that has multiple IP addresses for the same DLCI APARTXT NA02318 = Fix trace status for ip trace APARTXT NA02304 = Software primitive tftp does not work APARTXT NA02349 = CS failing link activation with parallel TG support inconsistancy sense code. APARTXT NA02310 = APPN does not come up after changing the TR MAC address APARTXT NA02305 = With connection to 2210, 6611 X.25 link recovery was not successful when cable was pulled then replugged. APARTXT NA02261 = Pacing on frame relay does not take into account, the number of saps and rips that are actually being broadcast. It doesn't consider Split Horizon or filtering that is configured. It just calculates based on the number of rips and saps in the tables. The result is that when big tables primarily learned from interfaces with split horizon cfgd are broadcast on. the rips and saps are bunched up in the first few pace intervals. With this fix the actual number of saps and rips to be transmitted is used to calculate the packet quantity per interval, so the packets are spread out over the whole interval. APARTXT NA02350 = architecture change to format of FR sap. APARTXT NA02263 = On recovery of PPP link, PPP establishes connection immediately but GATED / OSPF / IP, delay 75 seconds before IP makes the route available. Delay previously required because of problem with route flooding. Code was introduced earlier to resolve flooding problem and delay must now be removed. APARTXT NA02269 = Losing ability to start new DLSw sessions. Problem believed triggered by 2210 patch 44. 2210 closes a broken or fail-to- establish session in a way that causes 6611 not to remove dead entries from the message queue data base 9. Queues remaining have entries containing "NO_IP...". When database fills, with 200 of these entries, the analyst exits and no more sessions can be started with anyone. Problem also agravated by 2210 not haveing a DLSw partner defined as this causes the sessions to not establish. APARTXT NA02439 = Support MIA APARTXT NA02440 = Add APPN ping command to SysMan and fast-path APARTXT NA02441 = smds code not enabled without this feature APARTXT NA02442 = mtu sizes < 2k not supported APARTXT NA02443 = can NOT MIA on TR for bridging and IP. can NOT MIA or sigle port bridge and IP on Ethernet cards STOP_SYMPTOM APARTXT NA02300 = symptom not applicable. this is a feature. STOP_SYMPTOM APARTXT NA02185 = Add MIA support to SYSID screen APARTXT NA02487 = Implement SNMP changes for multi-ip addresses per interface APARTXT NA02444 = TIMED changes for multi-ip addresses per interface APARTXT NA02445 = SMDS migration to r4ap APARTXT NA02278 = Customer running NP00601 with FX60105. They start IPX traces and are unable to stop them so they grow without bounds. * We noticed /etc/fuser is not in this level of code APARTXT NA02353 = product program number and version release number incorrect APARTXT NA02446 = Link GateD 3.5 BGP-4/CIDR from v1r4 to r4ap. APARTXT NA02311 = If the router is brought down ungracefully(power switch, reset button, power outage) on 6611 models 175, 145, or 12x, when router is brought back up the date has moved one day forward. APARTXT NA02447 = Assertion failure in bgp_init.c around line 1376. APARTXT NA02314 = The button under transparent bridging to either filter or forward duplicate addresses has the inverse meaning. That is filter forwards and forward filters APARTXT NA02315 = Scram is dying because a stop interface is being issued by monintr to an interface that does not have a protocol address configured. The fix is to remove the check_kernel_state routine in monintr.c. SCRAM still shouldn't die under this condition but because the design effort required to fix this in SCRAM is significant and the check in monintr is redundant. APARTXT NA02448 = Fixes to gatedcfgd for the new Config Tool. APARTXT NA02449 = START_TEST Verify no break STOP_TEST APARTXT NA02450 = compile fix "not enough sram" APARTXT NA02316 = r4ap+ build did not have depenance on Make.cfg so when a change was made to include as a define SMDS_MODULE, normal makes failed to extract files that had SMDS_MODULE but did not build. Solution is to check these files out and make the following change #ifdef NOP #endif to these files APARTXT NA02313 = A MS DLS PC on an ethernet connection to 6611 has problems establishing a DLS session for SNA. DLS uses a test command to facilitate setting up its link station. MS DLS sends an XID command and refuses to answer the test command. The 6611 refuses to pass the XID command to the intended partner, so there is a deadlock situation. The architecture says that, both the Test command and the XID should be responded to immediately. Similar problem was fixed for Token Ring terminals already. APARTXT NA02451 = Make tcp_keepidle and tcp_keepintvl configurable as per the Fuji Bank requirement. APARTXT NA02322 = DLS does not respond with name recongized to second name query if second name query is within 10 seconds of first query. APARTXT NA02504 = Activating a prior level without the clock fix (22046) causes the clock to read 2/25/30. APARTXT NA02452 = GateD 3.5 supports non-disruptive time changes. sysmon.conf must be modified in order to exploit this capability. The link to /etc/sysmon.conf must be broken since GateD 3.5 is only shipped V1R4+. STOP_SYMPTOM APARTXT NA02354 = enhancement to SDLC management traps requested by Synovus APARTXT NA02453 = cfg-interface.start scripts do not include ptp_down IP needs to know link is not a point to point link in order for bcast/multicast messages to flow over SMDS link STOP_SYMPTOM APARTXT NA02325 = Add error msg to adapter dump when none selected APARTXT NA02326 = Peer-to-peer and free buffer list gets depleted in p385. THis problem was originally fixed in defect 17191, but it was zapped when the same file was changed in defect 18333 (DR code).This problem happens when there is a lot of intermixed multicast and multibuffer traffic going card-to-card (especially when ther are several multi-port adapters). APARTXT NA02454 = keep alive changes APARTXT NA02360 = no LLC_WRITE trace entry for ethernet APARTXT NA02329 = Software update apply FX60108 fails with message that FX60108 is not in transfer directory, but it really is there. Apply from sysman works. There is a problem with the fast path command not creating a new table of contents if one already exists. APARTXT NA02359 = merging product specific changes to go along with psna extract APARTXT NA02331 = /2210 when the serail link goes down then recovers the 6611 is not downloading the routes to the adapter. See Acee or Phuong. APARTXT NA02358 = PSNA defects missing in route66 APARTXT NA02373 = Need to change max_trees to 50 to match PVT/SVT testing APARTXT NA02503 = Internal Enhancements for NP00640 APARTXT NA02470 = Multicast packets on P2P interfaces sent with wrong source address. APARTXT NA02357 = Sending a config with changes that do not trigger a reboot, may sometimes disable some ports APARTXT NA02456 = none APARTXT NA02457 = xxxnos00 cannot be loaded APARTXT NA02458 = Enhancements for ipx for v1r4+: 1. Add pacing to ipx sap and rip packets on a ppp port. 2. Double the size of the ipx sap and rip tables to 4096 entries. APARTXT NA02379 = psna environment fixes missing APARTXT NA02364 = The symptoms of this problem are varied and include the possiblity of not being able to ping the broadcast address. Other symptoms include loss of contact with formerly reachable nodes. The problem usually is characterized by the Frame Relay arp table looking strange with multiple entries for identical IP addresses, no sometimes no broadcast entry at the end of the table and aiNumIp looking strange in comparison to the table contents. We found that the problem could be introduced when multiple DLCI dele tes were processed as a result of receiving an LMI message which indicate d that more than 1 DLCI was deleted or went inactive. Although our code correctly decremented the total number of entries in our Frame Relay arp table, it did not fully move all entries backwards as the table was shrunk in processing multiple deletes. APARTXT NA02460 = Configurable TCP Timers for Fuji Bank - new config object design. APARTXT NA02463 = Add a new dlstools option to reset a particular WAN connection. The connection will be identified by the IP Address of the DLSw partner on that WAN connection. APARTXT NA02374 = Have multiple routers configured with bridging in V1R4+ test bed. With V1R4, start up is done in parallel. Apparently two IFCONFIGs are hitting the same function at the same time and hanging. KLO says if another protocol were configured, it would rectify the situation - but only bridging is configured here. APARTXT NA02464 = BPDUS are not being sent out SR port APARTXT NA02375 = From the system manager try the following: Operations Network Statistics Bridge SRB Spanning tree statistics You get the following, instead of the interface names: /usr/bin/display_int [1430]:;ifnam:not found. APARTXT NA02465 = BPDUS are not being sent out SR port correct priority MASK is trunkateing actual value to 0. APARTXT NA02366 = When a PPP link goes Up/Down, a garbage ip addr. 186.223.202.17 appears in the neighbor table due to a mishandling of the LSA structure. Gated keeps logging find_nh_entry FAILED on that addr This problem only occurs at FX60105 and follow-on PTFs. APARTXT NA02376 = configuration fails when new card added. APARTXT NA02377 = configuration fails on interfaces > 9, ie. do10, do11. APARTXT NA02378 = APPN to a remote 6611 SDLC attached device does not work APARTXT NA02371 = sysmon filled with BAD write state 0 ?->? Local=(xxxxxxxxxx,f0) lanwrite.c--2190>>BAD write Sap f0 is now invalid, sysmon and error log fill with these messages, data link switch can no longer get anything through this adapter, but adapter is up, and you can ping through it. APARTXT NA02466 = Select Fixes from Rich Woundy: 1) Null route pointer check on interface structure deletion. 2) Correct check for direct route preference change required (direct routes changes to Intra-area route or vice-versa). 3) Add check for correct I_M_MS DB summary bits for FULL and LOADING states. APARTXT NA02467 = Gated will log an error if it receives a packet that has source address from different subnet. In MIA, an interface can have many different subnets and when they run OSPF using multicast address, a host can receive packets from different subnets. This creat problem because GateD logging error every time it sees a different subnet packet. APARTXT NA02372 = APPN out of shared storage, take appn dump, run fmtdata dump name > data.dump, search on Cells, observe large number of 16 by tes cells allocated, total storage used is above the warning threshold, or the critical threshold, no links or sessions can be started once above critical. APARTXT NA02468 = Remove references to library to avoid build team confusion. STOP_SYMPTOM START_TEST N/A - not part of EPTF. APARTXT NA02469 = Install this ptf on top of NP00624 and expect it to fail. APARTXT NA02381 = route setup failing because link in hpr cloud is actually not hpr capable. APARTXT NA02471 = cub winMIB interface does not produce output APARTXT NA02384 = Schedule config changed caused config deamon to dump. APARTXT NA02472 = As the analyst's message queue gets longer, the analyst spends more time getting messages off of its queue. This is because the analyst ALWAYS checks for a PRIORITY Msg before getting another message. This causes the system to look at the entire queue looking for a PRIORITY Msg when a majority of the time, one will not be there. This is very inefficient. A change has been made to send a signal to the analyst whenever a PRIORITY message is put to the analyst's queue. This signal will tell the analyst to get a PRIORITY Msg next time it gets a message. Otherwise, as in most cases, the analyst will just get the next message off the queue, bypassing the need to look at the entire queue. This will be much more efficient under heavy load (just when the processor needs it)! APARTXT NA02473 = OSPF_ASE routes with the same network but different masks are not supported. With this change they will be supported with the restriction that the advertisements must be originated by different routers. APARTXT NA02474 = The new function that was added for p640, called Serial Interface Leads Display, works as advertised from the FastPath, but fails from the SysMan. The error mesage from the SysMan is as follows: Interface: ct1 Type:6611 Serial/Token-Ring Combination Adapter Status: if_led[86]: sertype: not found APARTXT NA02510 = Incorrect logging of error when LNM linking is not completed. Remove logging. APARTXT NA02476 = formatted dump only shows ascii character representation; ebcdic should be shown as well APARTXT NA02477 = APING does not work from FastPath or SysMgr, though it does work normally from the back door. For instance, from the back door, in order to ping altair from arcturus, the following command line is used: aping -l ARCTURUS -f NETA.ALTAIR -v Note that the use of caps for the machine names is mandatory. Also, the prefix "NETA" that precedes the target machine's name is somtimes optional, but cannot be specified for the local machine. From the fast path, the following command line was used: appn ping ARCTURUS NETA.ALTAIR, which yeilds following error msg: /usr/bin/appn[2019]: aping: not found Just to be thorough, I tried every combination of including and excluding the "NETA" prefix on both the local and target machines. Ditto for the SysMgr. From the SysMgr, the following error message is displayed: ping_to_appn[68]: aping: not found APARTXT NA02478 = Problem with loop in findLSA with certain combinations of LSAs. APARTXT NA02479 = Gated on vega is not getting all the routes, that is, netstat -rn does not show all the routes that it should. This seems to be a rather specific problem, and does not happen all the time. vega's backbone port has to be down, and then reboot spica so that vega loses routes from the other side of the frame relay cloud. After spica is back up, vega should once again pick up these routes, but doesn't. vega has multiple routes to spica; if the configuration is simplified so that it has only a token ring and the frame relay ports, the problem does not occur. I have shown the problem to Acee. This is on PTF F064018. APARTXT NA02401 = APPN abnormally ended after CP-CP sessions were terminated and restar ted. Problem also occurs when DLUR sessions are terminated and restarted. The problem was identified by a F02B pexit code in the error log. Problem may have other symptoms resulting from having invalid internal APPN table entries. Problem was introduced for DLUR in base V1R4 and introduced for gener al APPN CP-CP sessions in V1R4 at NP00623. APARTXT NA02480 = This problem only happens on this box with X.25 shallow adapter. (However, it won't happen on another box with X.25 adpater... mystery!) Anyway, fix on SCRAM/IP not to exit when IP mask is not retrieved from kernel yet. ANother point: This problem only happens when reboot a router with X.25 adapter with different config diskette. It did not occur when reboot with the same config diskette! APARTXT NA02481 = psna load APARTXT NA02404 = When Scram somehow could not read nap_message sent from frame relay, it then closed the nap socket. Later on when Frame Relay sends nap messages to scram indicate that the DLCIs are UP, the message are discarded due to the closed pipe. As a consequence, the kernel interface structures were not updated, ospf therefore did not establish the ospf neighborhood. APARTXT NA02484 = Internal Enhancements for NP00640 APARTXT NA02483 = Add configurable retransmission count and retransmission interval as per a request from the Fuji Bank account. APARTXT NA02461 = I had a network of 17 routers using IP, OSPF, and BGP. After adding other protocols, Vines, IPX, XNS, and DECnet, gated seems to have started putting a lot of messages in the errorlogs on all the routers. The messages are: gated[]: KRT RECV MSG (null) router #3 do0 10.1a.0.0.11.10.2.ab.cd.ef.0.12.4.53.1.31 flags 43: ignoring. gated[]: KRT RECV MSG (null) router #6 te0 10.1a.0.1.92.10.aa.0.4.0.3.4.4.53.0.0 flags 43: ignoring. gated[]: KRT RECV MSG (null) router (null) flags 43: ignoring. These keep repeating, clogging the log. APARTXT NA02462 = The DLSw Capabilities Exchange function was coded to respond negatively to any Control Vector that it does not know about. This will make it very difficult for customers to migrate to newer versions of code where new Control Vectors exist. The code should be changed to just ignore all Control Vectors that are unknown to the 6611. STOP_SYMPTOM START_TEST Create a DLSw partnership between a 6611 and a 2210 running the latest level of code. This version has new Control Vectors. Check that the CapEx works and the partnership is established. APARTXT NA02485 = An unexplicably core dump was found. The dump was truncated due to the AIX limit for user processes. Truncated dumps aren't worth shit so code will be added to assure a full dump is generated if/when we get another. APARTXT NA02486 = Tiered netbios connection with RFC1795 fails APARTXT NA02426 = Integrity problem APARTXT NA02505 = Without this change, an ifconfig of an SMDS interface would have the Pointopoint rather than broadcast flag on and OSPF routes would not flow over the interface STOP_SYMPTOM APARTXT NA02506 = The problem is that lanwrite issues a FLOW CONTROL ioctl for a link station that has just gone down, thus the ioctl fails. Because of this failure, lanwrite calls HaltCircuitNow() which sends a HALTLS_NOW Msg over the WAN and then issues a DLC_HALT ioctl. Well, because the link station has already gone down this second ioctl fails also. And because it fails, the conversation state is NOT changed (which just happens to be REMDISCPEND in this case). Because the HALTLS_NOW was sent, we will never get a LS_HALTED Msg back, and thus we will remain in REMDISCPEND for the rest of eternity. APARTXT NA02525 = Config Mask Definitions need new value of "Both" for compatibility. APARTXT NA02531 = appn route test times out when issued from fast path or from sysman menus. I tried all 3 varieties available on the fast path, appn routes test -HPR_node_name NETA.SIRIUS.#INTER community appn routes test -number 1 community appn routes test 82:80 80:00:00:00:60:11:06:60 community APARTXT NA02526 = SMDS system manager options to display smds arp table did not work. Need to remove some options STOP_SYMPTOM APARTXT NA02527 = The command: appn routes test -number 20 appn gives the result: The NCE/TC connection number you specified, 20, is invalid The NCE/TC connection number must be between 1 and . It fails to list the upper bound. (20 was indeed too high in my case.) APARTXT NA02528 = SCRAM stops receiving NAP alert messages after reporting "No buffer available" hence DLCI up/down status is not received and OSPF neighbors are not established. (FUJI bank problem) APARTXT NA02529 = When you activate the PTF containing 22111 and then attempt to activate to a previous level, the operation fails to complete, and aborts in the shutdown phase. APARTXT NA02530 = Extra options and actions displayed for SMDS help. Also when display the smds arp table thru system manager operations, the contents of the table are correctly displayed but a command failed message shows on the system manager screen. STOP_SYMPTOM APARTXT NA02532 = The fix for defect 21032 was deleted from ip_func.c (version 1.11.1.24) by feature 13572. The problem by defect 21032 is a hit_and_miss error therefore it is NOT always detected by Eptf testing. I detected this missing when I helped Mike Sarisky (PE) on a field problem when a customer has a downlevel without this fix and I want accidently looked at ip_func.c (SCRAM) to make sure if it has this fix. As it turned out this fix was on ip_func.c (in R4AP) version 1.11.1.21, 1.11.1.22, 1.11.1.23 but not on 1.11.1.24. APARTXT NA02541 = Internal Enhancements for NP00640 APARTXT NA02488 = STOP_SYMPTOM APARTXT NA02489 = Add smds improvements APARTXT NA02490 = priority filtering STOP_SYMPTOM ENDSET COMPID FOR PTF NP00640 = 564801600,140 PRODID FOR PTF NP00640 = 5648016,Multiprotocol Network Program PREREQ FOR NP00640 = NONE ENDSET COREQ FOR NP00640 = NONE ENDSET SUP FOR NP00640 = NONE ENDSET IFREQ FOR NP00640 = NONE ENDSET MEMO TO USERS = +---------------------------------------------------------------+ | | | | | MEMO FOR NP00640 | | | | 640 01 | | | | | +---------------------------------------------------------------+ CONTENTS OF THIS MEMO _____________________ o Description of Supersede and Prerequisite PTFs o Installation instructions for NP00640 o Engineering PTFs with contents included in Production PTFs o Option set 1116 - NNS support for DLUR EN o BGP-4 o SMDS Integration o Multiple IP Addresses o IP Filters o IP Keep Alive o IP TCP Re-Transmission Control o Router Access via Bridged EN Port o APPN APING o IPX PACING AND LARGER SAP/RIP TABLES o Bridge Priority Queueing o System Manager Enhancements o Configuration Enhancements SUPERSEDE AND PREREQUISITE PTFS ________________________________ NP00640 is not a supersede PTF. There are no prerequisite PTFs for NP00640. Any user at base V1R4 can apply this ptf. Having PTFs NP00620 and NP00623, though not required, will not prevent NP00640 from being applied. However, a user having PTFs NP00625 or beyond will have to remove those PTFs before NP00640 can be applied. All the fixes contained in NP00623 and NP00620 are contained in NP00640. Users at any release levels other than V1R4 shall need to migrate to that level (using NP99931) and be committed at that level before applying NP00640. PTF INSTALL INSTRUCTIONS FOR NP00640 ____________________________________ Potential problems or delays in the installation procedure can be avoided if you perform certain actions first. CLEANING UP THE 6611 TRANSFER DIRECTORY Receiving installation files places them in the transfer directory of the 6611. If you know you have unnecessary files in the transfer directory (such as outdated dumps or reports), delete the files before beginning the installation. Refer to "Delete Transfer Directory Files" in the Operations chapter of the IBM Multiprotocol Network Program Operations and Problem Management manual for details. STOPPING TRACES Before beginning the installation process for any update, stop all trace activities. Active traces might cause installation to fail because of insufficient space. Refer to the Problem Determination chapter of the IBM Multiprotocol Network Program Operations and Problem Management manual for information on stopping traces. HANDLING DEVELOPMENT PTFS Before beginning the installation process for any software update, check to see if you have any development PTFs on your system. Development PTFs may be identified as containing the characters FX in their names, for example: Name: mpnp.obj NP00001 ---------------------- Fix ID Release ========================== FX00001 01.01.0000.0001 If you know that you have development PTFs on your system, call your IBM Support Center and provide the following information: o The development PTFs that are already on your system o The software updates you wish to install INSTALLING SOFTWARE UPDATES Software updates are installed from the transfer directory using the Software Installation and Maintenance Facility of the Multiprotocol Network Program V1R3 Supplement(SD35-0238). This is a multiphase process that permits the software updates to be tested before making them a permanent part of the Multiprotocol Network Program. When updating software, the first step after importing the software update is to apply the software update. After a software update is applied to a 6611, it can be activated at a later time. You can specify the maintenance level and the date and time of the activation. The 6611 will be re-IPLed at activation time. Thus, the application of a PTF is not directly coupled with the re-IPL of the router. Any software update that is applied, but not activated in the currently running 6611, can be rejected at any time. A software update must be active before it can be committed. Warning: When you commit the software update, the change is permanent. There are several rules to consider when installing software updates that have dependencies: o Software updates cannot be applied until their prerequisite software updates have been applied. o Applied software updates cannot be committed until their prerequisite software updates have been committed. Possible Messages When Applying PTFs: If there is not sufficient space available, you will see a system message similar to: Installation failed. Please press enter to view the detailed information. (Assuming Enter is pressed, this detailed message is displayed:) installp could not complete because there is not enough space in file system /usr. It needs 476 more 1K blocks. The space available in /usr is less than the minimum required for this operation. The selected install operation will not be started now. Press Enter to continue. If there are more than 1000 1K blocks free in each file system, then you will see different messages depending on whether there was sufficient room for the PTFs you selected to apply. NOTE: If you have insufficient space in the file systems other than /tmp and you have commited all prior PTF's, you must contact IBM service. IBM service will make the appropriate adjustments to the file systems. If you select multiple PTFs to be applied and there is room for some, but not all, only the ones that fit will be applied. The log file pd_update.log in the transfer directory will inform you of this. To perform software updates from the System Manager: 1. Log in to the System Manager using a controlling user ID. 2. Select Software Installation and Maintenance on the System Manager main menu. Follow this procedure to install your software update using the System Manager: Step 1 Your software update is sent to you on diskettes or tape. The software update must be moved to the transfer directory for installation. Before proceeding, refer to "Recommended Software Pre-Installation Actions" for suggested actions if you have not already done so. Step 2 Select Receive Software Updates from the Software Installation and Maintenance menu to move the update files to the transfer directory in the 6611. These import methods are available: * FTP * Modem * Tape * Diskette Before receiving any software files, the system validates that enough space is available in the transfer directory. NOTE: Steps 3 and 4 are optional steps for informational purposes only and are not required for this installation Step 3 Select List Software Updates in the Transfer Directory to view a list of the software in the transfer directory of the 6611, Step 4 Select List All Problems Fixed by Software Updates to list the problems fixed by the software updates in the transfer directory. Step 5 Select Apply Software Updates to install the software updates on the 6611. The installation ends without a re-ipl, and you are returned to the Software Installation Maintenance menu. Step 6 Select Activate Applied Software Updates to activate the applied software updates. In the first selection screen, select NP00640. In the second selector screen you must use "none" as a selection if that is the only choice or if you wish to select no EPTF's. COMMITTING THE SOFTWARE UPDATE (PTF NP00640) CAUTION: When you commit the software update, the change is permanent. You will not be able to reverse the commit. When PTF NP00640 is committed NP00620 and NP00623 if present WILL NOT BE automatically committed. They cannot, however, be activated, and can only be rejected, thereafter. Step 1 Select Commit Active Software Updates to commit the activated software updates. Refer to Chapter 5 "Software Installation and Maintenance" in the Multiprotocol Network Program Supplement Version 1 Release 3 (SD35-0238) more installation information. ------------------------------------------------------------ Installation of PTF NP00640 along with the prerequisite and corequisite PTF's, will alleviate the need for the engineering PTF's listed below to reside on your machine. NOTE: If you are currently running with an Engineering PTF that is not on the list below, contact your IBM Service representative for possible additional instructions. FX62001 FX62002 FX62301 FX62302 ----------------------------------------- NNS support for DLUR EN - Option Set 1116 ----------------------------------------- This feature enables the 6611, when configured as an APPN NN, to understand when a DLUR EN that supports Option Set 1116 registers its LUs. The Dependent LU Requestor (DLUR) architecture has been enhanced to allow DLUR End Nodes to register their LUs with a Network Node Server so that searches do not need to take place each time. This can cut down significantly on unneeded network searches. The 6611 cannot be configured as an End Node but now has the Network Node Server support to receive the LU registration and recognize the LU as a DLUR LU. ----------------------------------------- * BGP-4 ----------------------------------------- Border Gateway Protocol (BGP) uses policies to control the advertisement of routes across organization (autonomous system) boundaries. Thus BGP is used to interface an organization's network with one or more providers of Internet service. BGP is also used by service providers to interface with each other. Note that BGP can also be used to control the distribution of route topology information within a large private network. BGP-4 is described in RFC-1657. The key difference between it and BGP-3 has to do with aggregation of routes. With BGP-3 all advertisements must be for classical network addresses. BGP-4 allows contiguous addresses to be aggregated. Any address prefix can be advertised as long as the advertiser provides routes to all the included addresses. So the hope is that providers will be able to aggregate the network addresses of their users and reduce the large number of routes that must be advertised today for complete information about large networks. BPG-4 also facilitates CIDR (Classless Inter-Domain Routing) since it allows an organization to advertise a block of class C networks with a single advertisement. CIDR was shipped on the 6611 in V1R4. The 6611 implementation of BGP-4 has the following limitations: - It does not support the full Internet routing table (35-40,000 routes). - The BGP-4 MIB is supported as SNMPv1 rather than SNMPv2 as specified in the RFC. ----------------------------------------- * SMDS Integration ----------------------------------------- Switched Multimegabit Data Service is a high performance, connection-less, public switched data service designed to interconnect LANs and provides LAN-like features and performance across Wide Area Networks. SMDS uses cell relay switching technology. Generic system requirements are defined in Bellcore specification TR-TSV-000722. RFC-1209 defines the handling of IP and ARP packets and encapsulation, and SMDS complies with IEEE standard 802.6 MAN. The service provides individual, group and ARP addresses at DS1 (1.5Mbps), DS3 (45Mbps) and SONET (155Mbps) speeds. It also supports a slower speed option of 56-64Kbps. The 6611 supports speeds from 56Kbps through DS1. SMDS was initially shipped on the 6611 as a PRPQ (Programming Request for Price Quotation). In PTF NP00640, the code is incorporated into the mainstream 6611 MPNP code. The SMDS WAN protocol can be configured on A25 or A47 two or four port serial adapters or on either LAN/WAN combo adapter. It is not supported on the old serial adapters. **************************************************************** *SUPPLEMENT TO 6611 NETWORK PROCESSOR OPERATION POCKET GUIDE * **************************************************************** *** If you receive this document in softcopy format and *** *** print it, symbols, for example, in fast path commands, *** *** may not resolve accurately for some printers. Please *** *** refer to the IBM MPNP Operations and Problem Management *** *** Guide, SC31-6691 for a description of how to use these *** *** commands. *** **************************************************************** This chapter supplements the information in Chapter 4 of the 6611 Network Processor Operation Pocket Guide, GX27-3909. These guidelines describe how fast-path commands are structured: * Brackets ([ ]) surround: - Groups of options separated by an |, when only one option in the group can be specified (for example, [-diskette|-tape]) - Groups of options separated by an &|, when one or more options in the group can be specified (for example, [(-packet) &| -frame]) * Parameters are in double quotation marks (" "). * Parentheses (( )) surround any defaults for the subobjects, actions (depending on the subobject), and options. * Braces ({ }) surround options and parameters that are not required. ******************************************************************** SMDS Commands Use the SMDS commands to: * View the SMDS ARP Table * Delete all entries from the SMDS ARP Table * Clear SMDS statistical counters * View the SMDS individual address, group address and ARP address associated with a given interface * View the SMDS statistics for a given interface * List interfaces configured for SMDS * View the SMDS MIB Error table for a given interface Object Abbreviations: smds: SMDS, sm Subobject Abbreviations: address: addr, ad arp: a statistics: statistic, stat, s Action Abbreviations: clear: clr, c delete: del, d list: l view: v Option Abbreviations: -active: -act, -a -all -miberr: -me Summary of SMDS Commands smds address (view){-log} "interface" - View the SMDS individual address, group address and ARP address associated with a given interface. smds arp delete (-all) "interface" - Delete entries from the SMDS ARP Table. smds arp (view) {-log} "interface" - View the SMDS ARP Table unsorted. smds list {-active} {-log} "interface" - List the serial interfaces configured for SMDS. smds statistics clear "interface" - Clear SMDS statistical counters. smds statistics (view) {-log} "interface" - View the SMDS statistics for a given interface. smds statistics (view) -miberr {-log} "interface" - View the SMDS MIB Error table for a given interface. ----------------------------------------- * Multiple IP Addresses (MIA) ----------------------------------------- Prior to this function, the 6611 supported only one IP address per physical interface. This function supports multiple IP addresses on single token ring and ethernet ports. The 6611 implementation of MIA has the following limitations: - MIA is only supported on LAN adapters (token ring and ethernet). - A maximum of 10 IP addresses is allowed per physical interface with a total of 66 addresses per router. The network (sub-network) portion of the secondary address of each interface must be unique within the router. - Each IP address may have its own subnet mask. - No subnet overlapping is permitted. This restriction does not apply to the loop-back address (127.0.0.1) or the multicast address (224.0.0.x). ----------------------------------------- * IP Filters ----------------------------------------- This function allows a customer to enable traffic such as telnet and ftp (both requests and responses) to flow in one direction but not in the other. An example would be where a user might want to telnet/ftp out to the Internet but where the customer might want to restrict external access into his/her network. Customers can now choose filtering based on TCP/UDP source port, destination port or both. Prior to NP00640, filtering had to be symmetrical, that is, the restrictions had to be the same in both directions. ----------------------------------------- * IP Keep Alive ----------------------------------------- The IP "keep alive" timeout period can now be tuned by the customer to enhance the stability of some networks. Prior to PTF NP00640, this value could not be altered. ----------------------------------------- * IP TCP Re-Transmission Control ----------------------------------------- Prior to NP00640, the 6611 always attempted 12 retransmissions prior to closing a TCP session. The maximum time between retransmissions was fixed at 64 seconds. With this function, the customer can alter these values via the Configuration Tool to provide better network stability in certain situations. ----------------------------------------- * Router Access via Bridged EN Port ----------------------------------------- A bridged Ethernet port can be accessed via telnet for the purposes of logging into the box. Prior to this PTF, this function was only available on token ring ports. ----------------------------------------- * SDLC MIB ----------------------------------------- Ther are two new SNMP SDLC traps that provide port and link station (LS) status. Port UP/DOWN status has also been made more accurate. Note that this function was also provided in V1R4 via PTF NP00623 and in V1R3 via PTF NP00387. ----------------------------------------- * APPN APING ----------------------------------------- An "APING" can be issued from a 6611 to any other APPN device supporting APING including another 6611, VTAM or Comm Manager/Server. APING may also be issued from one of these latter devices to a 6611. The target device does not have to be physically adjacent to the originator as long as an APPN "route" is active between the two devices. APING can be issued from a 6611 via the System Manager fastpath or a via a System Manager screen. ----------------------------------------- * IPX Pacing and Larger SAP/RIP Tables ----------------------------------------- IPX has been enhanced to provide SAP and RIP tables with 4096 entries which is twice the size of the tables in V1R4 and prior releases. In addition, IPX SAP and RIP pacing has been added to the PPP interface. ----------------------------------------- * Bridge Priority Queueing ----------------------------------------- This function provides bridge filtering that can be applied to determine the priority of SEND frames as high/medium/low. ----------------------------------------- * System Manager Enhancements ----------------------------------------- The System Manager has been enhanced to provide screen and fastpath commands for APPN network debugging and other functions. Note that APPN fastpath commands were also provided in V1R4 PTF NP00623. ----------------------------------------- * Configuration Tool ----------------------------------------- The 6611 Configuration Tool has been enhanced to support the new functions in PTF NP00640. This Tool may also be used to configure 6611 V1R4 routers. A Configuration UPGRADE is NOT required. The customer is warned if he/she attempts to configure function only available in PTF NP00640. In most cases, if the customer chooses to ignore this warning and configures enhanced function on a V1R4 router without PTF 640, the new function will be ignored -- however, this is not guaranteed. Additional information specific to the new Configuration Tool can be found in the README file on the Configuration Tool installation diskettes. END OF MEMO - NP00640 ENDSET LABELTXT Selective Fix Package For use on MPNP V1, R4.0 only ENDSET ***************************************************