DECnet config problem

[geir_2]

Hi,

I have recently configured two VMS nodes with following address: 1.81 and 2.81.

VMS nodes is separate by a Cisco WAN connection (configured as a IP-tunnel)

VMS-A (1.81)--CiscoA ==== CiscoB--VMS-B(2.81)

It's possible to use ping and set host command locally, adn remotelyf between the cisco router. But if I try to use the set host command fram VMS-B to VMS-A the remote area is not reachable.

The DECnet routing table on the cisco router looks correct.

Maybe some DECnet commands is missing on the VMS host. When I run the following ncp commands on VMS-B

ncp > show active area
I can't see area 1, onlut area 1 is available.

I hope someone could help me :-)

Thanks Geir

[Volker Halle]

Geir,

are these Cisco Routers running DECnet Phase IV Area routing ? There must be an area router in area 1 and another area router in area 2 for the 2 VMS nodes to be able to establish connections via DECnet.

NCP SHOW KNO NODES on the VMS systems should show adjacent routers. NCP SHOW KNO AREA should show known area routers.

Volker.

[geir_2]

Hi Volker,

When use following commands, I had connection between VMS-A and VMS-B on IP.

ucx> set route/default/gateway=10.2.1.1

But when I reboot the computer, the information is lost. So I must save the information permanently, but I did not know how it's done.

The VMS-B is not able to see other area, and nodes. I must have forgotten some commands when I configure the DECnet on the host.

VMS-B can use set host 0 and set host to 2.81 (which is VMS-B's address)




Hope someone could help me.

[Volker Halle]

Geir,

DECnet and IP are completely different protocols and use different routing mechanisms. If you would be using DECnet-over-IP, DECnet packets would travel over the TCP/IP route, but you seem to be running DECnet Phase IV.

To create a permanent route entry for TCPIP, use

UCX> set route/default/gateway=10.2.1.1 /PERM

To be able to reach another DECnet area, a DECnet node must have a path via an area router in it's own area and an area router in the destination area to get to a node in the destination area.

What area the DECnet addresses for the two Cisco Routers ?

Volker.

[Hoff]

You need a DECnet level two (area) router in area one and a second DECnet area router in area two, and both must be located somewhere between these two nodes -- if the Cisco switch connection is implemented per norms, the tunnel is transparent and not relevant -- and only with these two adjacent area routers can a node (either end-node or level one router) in DECnet area one communicate with a DECnet node (end-node or level one router) in area two.

Or you can move the nodes into the same area, and use a level one (non-area) DECnet router, or -- with no router available -- the ability of end-nodes to communicate directly on a LAN. This assumes the Cisco provides a transparent LAN connection. Adding an external DECnet router (level one) or switching one or both of the nodes to a router is recommended, as it allows better control and management of the DECnet traffic. Many Cisco routers can be enabled as DECnet routers.

Or you can switch both of the nodes involved into area routers.

Or you might be able to set up the Cisco switches as DECnet area routers, and switch over from a tunnel/VPN over to a DECnet routed (inter-area) link for DECnet traffic.

The ability to ping nodes is not directly relevant to DECnet communications -- well, other than its potential use as a DECnet transport for DECnet-Plus (DECnet Phase V) hosts. But when using DECnet over IP, you still need to follow the DECnet rules.

DECnet phase IV (level one and level two area) routing is covered in some detail in the DECnet architecture manuals, and these are available via pointers in the OpenVMS FAQ. The DECnet-Plus user and system management manuals are available over at the HP OpenVMS documentation web site. The Phase IV manuals are probably posted somewhere, but they're not immediately available via the HP OpenVMS documentation web site, either directly or in the archived documentation section.

Stephen Hoffman
HoffmanLabs

[geir_2]

Hi,
Thanks for the answers :-). Yes, both router is configured as a Area-Router with following address 1.1 and 2.1.

Inside these router I use 1.81 and 2.81, which is configured as a non-routing device. (end-node)

I discovered yesterday that both VMS node used the same scsnodeid, which is configured in sysgen. Now I configured the correct scsnodeid on both nodes, which should be Area x 1024 + Nodeid.

The decnet routing table seems to be correct on both routers.

I have also rebooted one of the VMS nodes, later today I will try to rebbot the other nodes. So far no success.

[geir_2]

Hi,

If I ran following commands on VMS-B I got following respons:

NCP> show char area 2

Area Volatile Characteristic as 23-jan-2007 09:14:33

Area 2
State = Unreachable

Regards Geir


PS:
It's possible to use set host 0 and set host VMS-B, but not to VMS-A remotely. Hope someone could give me some input/hints ;-)

[Colin Butcher]

Hello Geir,

You imply that the Cisco's are running DECnet routing over the "IP tunnel". Later on you imply that you have "DECnet over IP" active (ucx set route).

Please can you specify the exact versions of VMS, DECnet, TCP/IP and the precise configuration. A diagram would help a lot.

Assuming that you are using DECnet Phase IV then you will need to set up DECnet Phase IV routing as an area router in the two Cisco routers. Alternatively you might configure the Ciscos to bridge non-IP traffic between the sites (this would also allow LAT and other layer 2 protocols through as well).

If you are using DECnet Phase V (DECnet-Plus) then you can use Phase IV compatible addressing with the Ciscos configured as above. Alternatively you can set up "DECnet over IP" and thus avoid the need for the Ciscos to do anything other than IP routing in the usual manner.

It depends what level of availability, multiple path operation, load balancing and so on you wish to achieve over the WAN. The easiest configuration these days is probably to use "DECnet over IP" and have minimal setup in the Cisco routers, however that may not achieve what you need, especially if there are other protocols as well, such as LAT (a layer 2 non-routable protocol).

You might find this of use: http://h71000.www7.hp.com/openvms/journal/v5/index.html#decnet for background reading.

Hope this helps. Cheers, Colin.

[geir_2]

Hi,

Thanks for the answer. I will try to draw a diagram of the network + VMS-nodes

Vms-A(1.81) SCSNODE=VMS-A,SCSSYSTEID=(1105)
|
CiscoA(1.1)
||
||
-DECnet is transported inside IP
||
||
CiscoB(2.1)
|
VMS-B(2.81) SCSNODE=VMS-B, SYSTEMID0(2129)


IT's full connectivity between router and hosts, except VMS-B.

Some history:
I took image backup of VMS-A, and restored it on VMS-B, and later I changed the name and address.

Im'm running DECnet OpenVMS AXP V6.2, and management version is 4.0. The HW is Alphaserver 1000.

Regards geir

[Volker Halle]

Geir,

try the following from VMS-A:

$ MC NCP TELL 1.81 SHOW EXEC
$ MC NCP TELL 1.1 SHOW EXEC
$ MC NCP TELL 2.1 SHOW EXEC
$ MC NCP TELL 2.81 SHOW EXEC

I don't know, whether the Cisco router would actually display any usable executor characteristcs data, but you should NOT get a %SYSTEM-F-UNREACHABLE error !

Then you can do the same kind of NCP SHOW commands from node VMS-B.

This will tell you the DECnet reachability of all the components involed in your little DECnet multi-area network.

Volker.

[Hoff]

The Cisco widgets are either not configured as DECnet area routers, or not directly connected (adjacent), or the Cisco router devices are not functional, or there is something between the end-node DECnet hosts here and the Cisco DECnet routers that is blocking the protocol traffic between the end nodes to the area routers.

Area routers are the only path between two DECnet areas, and area routers must have adjacent connections with other area routers. Connections between area routers cannot be routed through level one routers. (Well, DECnet area routers can be connected that way, but that's a configuration for another discussion and for another application entirely.)

Is there a reason you have DECnet nodes in two areas and (apparently) no functional area routers? That's a configuration that's rather more complex than usually necessary.

What to do? Examine each DECnet device. Determine if it is configured as required. Ensure the Cisco routers are area routers; level two DECnet routers. Also determine if you have duplicate DECnet addresses -- non-trivial Cisco environments involving virtual LANs and protocol bridging can easily have loops in the network topology, and all manner of weirdness can ensue. If DECnet itself detects a duplicate, you'll get an IVADDR message during DECnet startup. And consider moving to one area with one (or maybe two) level one routers, or with a virtual network.


As for displays:

End-nodes cannot see what's in the area, only the routers can.

End-nodes can display nodes that they know about locally.

Level one routers know about the nodes that are reachable within the router's level one area.

NCP commands commonly used here include SHOW EXEC CHAR and TELL node SHOW EXEC CHAR. You may need to add the requisite username and password specifications to access the remote nodes. You can also display adjacencies, and the EXEC CHAR displays will show if the node is an end-node, level one router, or level two (area) router. You'll also see here if the node is running DECnet Phase IV or DECnet Phase V (DECnet-Plus), as the latter has quite different management and can use different routing.

[geir_2]

Hi,

Thanks for the answers. I was not at office last week, It was impossible to answer, and test the proposals. But so far no results.

I discovered on VMS-B that active links was not active. Is the any commands I can use if I want to activate the links and delay.

[Volker Halle]

Geir,

on VMS-B try:

$ MC NCP SHOW EXEC CHAR
$ MC NCP SHOW ACT LINES
$ MC NCP SHOW ACT CIRCUITS

If there is no active circuit (or line), you may not have a valid DECnet license or your configuration is wrong, try:

$ MC NCP SET KNOWN LINES ALL
$ MC NCP SET KNOWN CIRCUITS ALL

and watch for error messages.

Volker.

[geir_2]

Hi,

I tried the commands, with no error messages. If I tried

mc ncp> show acti circ

I got followings messages:

EWA-0 on -synchronizing

Any comments??

Geir

[Volker Halle]

Geir,

there may be a physical problem with the LAN interface (EWA) or it's connection to the switch. Check the counters with MC NCP SHOW LINE EWA-0 COUNTERS

You may also try to clear the re-load the DECnet circuit:

$ MC NCP SET CIR EWA-0 STA OFF
$ MC NCP CLEAR CIR EWA-0 ALL
$ MC NCP SET CIRC EWA-0 ALL

Volker.

[Volker Halle]

Apparently continued in:

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=1094905

Volker.

[geir_2]

Volker,

If I removed the cable, the ping command doesn't get any aswer from remote nodes/hosts.

Following licenses is installed on the AlphaServer NITFS1: DVNETEXT and DVNETEND. It's the same as FBUFS1. The license is different(other Authorization number at each sites.

If I use following commands I got followings messages/output

$ show network

Node Links Cost Hops Next hop to Node
2.81 NITFS1 2 0 0 (Local) -> 2.81 NITFS1
2.81 NITFS1 2 0 0 (Local) -> 2.81 NITFS1
2.81 NITFS1 2 0 0 (Local) -> 2.81 NITFS1
2.81 NITFS1 2 0 0 (Local) -> 2.81 NITFS1
....
.....
2.81 NITFS1 2 0 0 (Local) -> 2.81 NITFS1
2.81 NITFS1 2 0 0 (Local) -> 2.81 NITFS1


Looks like a loop. But I dont' know why
Any comments??? Hope so :-)

[Hoff]

I've certainly seen various loops and whorls within Cisco networks. Ask the local networking folks or the Cisco support folks to look at this network, and about what nodes are located on what VLANs, and at where the packets are going. I would also look at the DECnet adjacencies in the network, as part of this.

"Check your Cisco routing. Most Cisco devices can be set up to route DECnet, or to bridge it; to treat it as its own virtual LAN. You'll need to use the Cisco monitoring to determine what IP and DECnet address(es) are where, and whether this is a routed or bridged configuration."

"This would not be the first Cisco DECnet configuration with routing and bridging/VLAN oddities."

[geir_2]

Hi,

Thanks for the answer. I have configured a lot of cisco router with DECnet routing during the last years (without problems)

I can see that the cisco router send out hello Level 2 packet to the ethernet and tunnel interface. The router also receive level 2 packet from NITFs1 and tunnel interface as well.

Maybe I should change type from area to non-routing on NITFS1.

So far a I can see, it must be a VMS problem.

The Router IOS and config is identical at each site. The OpenVMS and DECnet version is identical at each site. I have installed correct license on both VMS nodes.

It must be some NCP commands that solve the problems


Regards Geir :-)

PS:
Maybe i should run a new restore on NITFS1??