topic Re: LANCP connects, then fails in Operating System - OpenVMS

LANCP connects, then fails

Andreas Freiherr — Wed, 11 Mar 2009 15:24:01 GMT

After using DECnet (phase IV) MOP successfully for decades, I want to move to LANCP now for a couple of trusty old DECservers. I think I have everything set up properly for DLL and console connections, but I have only partial success:

LANCP> conn node tshn01/dev=ewa0
Connecting to 08-00-2B-05-9D-6A via EWA0 ....
MOP V3 format selected (1534 cmd size), type Control-D to disconnect
#
DECserver 200 Terminal Server V3.3A (BL40) - LAT V5.1
Raum ARES XMZ W251

Please type HELP if you need assistance

Enter username>
%LANCP-E-NORSP, No response from target node, device EWA0
LANCP> Fr
%CLI-W-IVVERB, unrecognized command verb - check validity and spelling
\FR\
LANCP>

As you can see, the DECserver greets and prompts me for a username, so there IS a connection. However, while I try to enter "Fr" as the username, the connection fails, and the input goes to LANCP's command processor instead.

This happens with OpenVMS V8.3 on a DS20E and is reproducible except for a slight variation in timing: sometimes, the connection drops before I can enter the terminal server's access password in response to the "#" prompt, sometimes it lasts a little longer, but I have not yet been able to issue any DECserver commands before I got disconnected.

I tried another DECserver with the same result.

I installed patch VMS83A_LAN-V0300, which -- according to its release notes -- is supposed to improve robustness of LANCP CONNECTs, but the symptoms remain the same.

Unfortunately, neither HELP/MESSAGE nor Google turned up any information about the %LANCP-E-NORSP message, so I could use some hints on how to proceed with troubleshooting.

Does it mean anything if MCR NCP CON VIA EWA-0 PHY ADDR... still delivers a reliable connection? (DECnet MOP is turned off if SHOW CIRC EWA-0 CHAR says "Service = disabled", correct?)

Thanks in advance for any and all help!
Andreas

Re: LANCP connects, then fails

Hoff — Wed, 11 Mar 2009 16:15:52 GMT

Try turning off DECnet MOP service; I'd expect that only one stack can have MOP lit up at a time.

Re: LANCP connects, then fails

Andreas Freiherr — Wed, 11 Mar 2009 16:22:01 GMT

Hoff,

Thanks for the hint. I'd agree there may be interference if MOP is on in both LANCP and NCP at the same time, but doesn't this "Service = disabled" mean that it's off here?

NCP>sh ci ewa-0 c

Circuit Volatile Characteristics as of 11-MAR-2009 18:20:41

Circuit = EWA-0

State = on
Service = disabled
Designated router = 1.244 (R1HNEQ)
Cost = 4
Maximum routers allowed = 33
Router priority = 64
Hello timer = 15
Type = Ethernet
Adjacent node = 1.244 (R1HNEQ)
Listen timer = 45

NCP>

Regards,
Andreas

Re: LANCP connects, then fails

Hoff — Wed, 11 Mar 2009 16:57:19 GMT

I don't see a reason for the connect to fail.

DECnet MOP service is correctly disabled in Phase IV here, yes.

There's a replacement LANCP image in various of the V8.3 UPDATE kits. Are you current on your (other) ECO kits?

That # is reminiscent of the server password knob.

Here's the template of what I usually use:

http://64.223.189.234/node/233

Re: LANCP connects, then fails

Richard Stockdale — Wed, 11 Mar 2009 22:16:36 GMT

Not much has been done with LANCP CONNECT in recent years, until a few months ago, when some recent support for LANCP CONNECT/INPUT=file had some problems. So a lot of work was put into LANCP CONNECT over the last few months and the new image might resolve your problem. This will be in the next VMS83*LAN patch kits, which should be out in a few months. If you can escalate the problem you can get a new LANCP much sooner.

- Dick

Re: LANCP connects, then fails

Wim Van den Wyngaert — Thu, 12 Mar 2009 09:17:42 GMT

Long time ago I had simular problems. The MOP of DECnet and LANCP are not the same.
They react differently on transmission problems. In my case, the download of the image failed partly(not always). The decserver had a full duplex line instead of a half duplex. Solved that and both mops where working fine.

Wim

Re: LANCP connects, then fails

Andreas Freiherr — Thu, 12 Mar 2009 10:04:04 GMT

Thanks, all of your hints are really appreciated!

Hoff: your site was one of the sources that I consulted to find guidelines for setting up LANCP. Some cookies, but then, useful information; earned you a prominent place in my bookmarks!

Not sure what other ECOs might be required. Here's what I currently have:

HN05$ pro sh hi
------------------------------------ ----------- ----------- --- -----------
PRODUCT KIT TYPE OPERATION VAL DATE
------------------------------------ ----------- ----------- --- -----------
DEC AXPVMS VMS83A_LAN V3.0 Patch Install Val 10-MAR-2009
DEC AXPVMS VMS83A_ACRTL V5.0 Patch Install Val 04-MAR-2009
LGTO AXPVMS SNCLNT732_ECO7 V7.3-2 Patch Install (U) 20-FEB-2009
LGTO AXPVMS SNCLNT V7.3-2 Full LP Install (U) 20-FEB-2009
DEC AXPVMS COBOL V2.9-1453 Full LP Install Val 19-DEC-2008
DEC AXPVMS VMS83A_RMS V9.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_F11X V1.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_FIBRE_SCSI V5.0 Patch Install Val 19-DEC-2008
DEC AXPVMS V83_ACMELDAP_STD V1.3 Patch Install (U) 19-DEC-2008
DEC AXPVMS V83_ACMELOGIN V1.1 Patch Install (U) 19-DEC-2008
DEC AXPVMS VMS83A_SORT32 V1.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_MONTOR V2.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_IMGDMP V1.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_HYPERSORT V1.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_COPY V2.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_BASRTL V2.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_UPDATE V7.0 Patch Install Val 19-DEC-2008
DEC AXPVMS VMS83A_PCSI V2.0 Patch Install Val 19-DEC-2008
HP AXPVMS KERBEROS V3.1-152 Full LP Install Val 19-DEC-2008
HP AXPVMS KERBEROS V3.0-103 Full LP Remove - 19-DEC-2008
CPQ AXPVMS CDSA V2.2-271 Full LP Install (C) 11-DEC-2008
DEC AXPVMS DECNET_PHASE_IV V8.3 Full LP Install (U) 11-DEC-2008
DEC AXPVMS DWMOTIF V1.6 Full LP Install (C) 11-DEC-2008
DEC AXPVMS DWMOTIF_SUPPORT V8.3 Full LP Install (U) 11-DEC-2008
DEC AXPVMS OPENVMS V8.3 Platform Install (U) 11-DEC-2008
DEC AXPVMS TCPIP V5.6-9 Full LP Install (C) 11-DEC-2008
DEC AXPVMS VMS V8.3 Oper System Install (U) 11-DEC-2008
HP AXPVMS AVAIL_MAN_BASE V8.3 Full LP Install (U) 11-DEC-2008
HP AXPVMS KERBEROS V3.0-103 Full LP Install (C) 11-DEC-2008
HP AXPVMS SSL V1.3-281 Full LP Install (C) 11-DEC-2008
HP AXPVMS TDC_RT V2.2-107 Full LP Install (C) 11-DEC-2008
------------------------------------ ----------- ----------- --- -----------
31 items found

Dick: OK, so I'll stay tuned for future ECOs. Meanwhile, I can just revert to NCP for the console connections. It would/will be nice, however, to use the node name and LANCP's database instead of NCP's PHYSICAL ADDRESS to CONNECT. This is reason enough to check back for patches.

Wim: Yes, been there, too. We had serious unexplained performance issues with our LAN a couple of years ago until we found out that we had lots of duplex mismatches. Solved that, and performance improved drastically for affected nodes. The DECserver's port looks like this (it's on a Cisco 2950):

interface FastEthernet0/48
duplex half
speed 10
spanning-tree portfast

which should be fine to the best of my knowledge. (Also, the HEARTBEAT characteristic of the DECserver is disabled to match the setting of the SQE switch at the transceiver.)

Re: LANCP connects, then fails

Volker Halle — Fri, 13 Mar 2009 06:19:29 GMT

Andreas,

It would/will be nice, however, to use the node name and LANCP's database instead of NCP's PHYSICAL ADDRESS to CONNECT.

Just define the node in the DECnet database:

$ MC NCP DEF NODE xyz HARDWARE ADDRES 08-00-2b-xx-yy-zz [CIRCUIT xxx-0]
$ MC NCP SET NODE xyz ALL
$ MC NCP CONN NODE xyz

Volker.

Re: LANCP connects, then fails

Wim Van den Wyngaert — Fri, 13 Mar 2009 07:21:44 GMT

Did you try show server count on the decserver to see if something is incorrect ?

Wim

Re: LANCP connects, then fails

Wim Van den Wyngaert — Fri, 13 Mar 2009 07:28:03 GMT

I know nothin but found this.

Spanning tree PortFast is an enhancement that we use to speed things up in a new deviceâ s introduction to the network. When a port comes online, it needs to figure out where it fits within the existing spanning tree. It does this through BPDU information, but spanning tree goes through multiple stages. It blocks for 20 seconds, listens for 15 seconds and learns for another 15 seconds. Then, if everything is good, itâ s placed in a forwarding state.

During the previous 50 seconds, however, no data is being passed on the link. This means many things like DHCP discovery, Windows logins, etc. may not work for workstation devices. Users tend to be irked by this behavior. This is why we use PortFast. It â jumpsâ to a forwarding state immediately because you have assured the switch that no loop will exist through this port.

Wim

Re: LANCP connects, then fails

Volker Halle — Fri, 13 Mar 2009 08:24:28 GMT

Andrease,

once LANCP has established the 'connection' and reserved the console port on the DECserver, it enters a little loop:

- read terminal input, check for disconnect char
- send input data to DECserver
- wait and read DECserver response
- output received data from DECserver

If it fails to receive a DECserver response within about 1 second, it will abort with the NORSP status.

You could try to trace the LAN traffic between the OpenVMS node and the DECserver to determine, what may be happening.

Volker.

Re: LANCP connects, then fails

Andreas Freiherr — Thu, 19 Mar 2009 10:07:33 GMT

Thanks again for taking the time and consideration, Volker and Wim!

Volker: it's a good idea to add the node to the DECnet database. I didn't think of that because the DECserver does not have a DECnet address, and the CIRCUIT parameter apparently cannot be stored, but if I add the VIA option to the CONNECT command, it works well:

NCP>def node tshn01 hard addr 08-00-2B-05-9D-6A circ ewa-0
%NCP-W-PRMNA, Parameter not applicable , Circuit

NCP>def node tshn01 hard addr 08-00-2B-05-9D-6A
NCP>set node tshn01 all
NCP>con node tshn01
Target does not respond
NCP>con node tshn01 via ewa-0
Console connected (press CTRL/D when finished)

#

... so it's a very workable replacement for LANCP's connect until new ECOs are available.

Tracing the packets would certainly what I'd do next if I were still at the university, but I cannot take the time for that in my current position. Sigh! ;-)

Wim: of the counters in the right half of the SHOW SERVER COUNTERS display, only "unrecognized destination" changes, whether CONNECTs fail or not. I seem to remember that more than 10/sec can slow down the DECserver, we are below that limit.

The documentation that you found about the "portfast" is exactly why we have that characteristic enabled: basically, we tell the switch "don't think twice, it's alright" and make the port come up faster after the physical connection is made, by skipping the spanning tree algorithm. For a DECserver, this could, for example, mean that the first download attempt succeeds and we need not wait for a retry. The downside of doing so is that loops in the cabling (one switch port wired to another in a circle) can lead to packet storms.

Thanks again!
Andreas

Re: LANCP connects, then fails

Volker Halle — Fri, 20 Mar 2009 05:39:47 GMT

Andreas,

NCP>def node tshn01 hard addr 08-00-2B-05-9D-6A circ ewa-0
%NCP-W-PRMNA, Parameter not applicable , Circuit

If you want to/need to define a 'service circuit' for that node, you need to use the correct syntax:

NCP> DEF NODE xxx SERVICE CIRCUIT ewa-0

But this would only be necessary, if you have multiple DECnet Circuits with MOP service enabled and want to load/connect to this node via a specific service circuit.

Volker.

Re: LANCP connects, then fails

Wim Van den Wyngaert — Fri, 20 Mar 2009 06:42:47 GMT

fwiw I have decservers with unrecognized destination increasing with 1 per second.
Others have 0. Both work fine (only console management and access via TSM).

Wim

Re: LANCP connects, then fails

Andreas Freiherr — Fri, 20 Mar 2009 14:03:51 GMT

Thanks, Volker!

After specifying the SERVICE CIRCUIT (by actually including the word SERVICE ;-) ) in the DECnet database, I can now simply CONNECT NODE xyz, without the need to specify any technical details about the actual connection, like a PHYSICAL ADDRESS or the interface VIA which the path leads to the DECserver.

I will revert to using good old NCP until further notice and maybe try LANCP again after applying future ECOs. My "unrecognized destination" counters are ticking faster than Wim's, which I understand as a manifestation of a larger network with louder background noise: we have several hundreds of PCs in a switched LAN here. Maybe LANCP's implementation of the console carrier protocol in it's currently released version is more susceptible to environmental pollution than NCP's.

Thanks again for taking your time and sharing your thoughts!

Andreas