Decnet Over IP

[Himanshu_3]

Hi All,

I am having an issue connecting my standalone OpenVMS 6.2 to another VMS machine on the network.

There has been one router involved in between my node and the other remote node.My network guys say that they have been able to decnet ping my machine and the other machine from their router. I suppose they are mentioning some inbuilt cisco decnet ping utility.

I have been told that the router converts the decnet messages from my machine and encapsulates them into decnet over ip. Does it mean that I do not have to configure decnet over ip on my machine ?

I have configured the decnet on my machine with net$config advanced and also given the ports 339 to be used.But it seems that these ports are not been used as the decnet over ip conversion is done on router not machine.

I am not able to do set host to the remote machine although my machine is "decnet pingable" from router.

Any suggestions please ?

Regards,
Himanshu

[Himanshu_3]

Here is some more info about the config

NCL>sho imp

Node 0
at 2005-10-09-07:17:54.173+01:00Iinf

Characteristics

Implementation =
{
[
Name = OpenVMS AXP ,
Version = "V6.2 "
] ,
[
Name = DECnet/OSI for OpenVMS ,
Version = "DECnet/OSI for OpenVMS Version V6.3 21-OCT-1995 06:17:34.25"
]
}

NCL>

NCL>show session control all

Node 0 Session Control
at 2005-10-09-07:18:38.103+01:00Iinf

Status

State = On
UID = EC7366AE-3768-11DA-8001-AA0004006618
Backtranslation Directory =

Characteristics

Incoming Timer = 45
Outgoing Timer = 60
Outgoing Proxy = True
Incoming Proxy = True
Address Update Interval = 0
Update Retry Interval = 60
Version = V3.0.0
Non Privileged User =
Modify ACS = True
Maintain Backward Soft Links = True
Node Synonym Directory =
Soft Link Timer =
Transport Precedence =
{
}
Naming Search Path =
{
[
Directory Service = Local ,
Template = "*"
] ,
[
Directory Service = Local ,
Template = "local:*"
] ,
[
Directory Service = Local ,
Template = "local:.*"
]
}
Backtranslation Search Path =
{
[
Directory Service = Local ,
Template = ""
]
}
Naming Cache Timeout = +30-00:00:00.000Iinf
Naming Cache Checkpoint Interval = +0-08:00:00.000Iinf

Counters

Creation Time = 2005-10-07-20:31:14.396+01:00Iinf
Access Control Violations = 0
Backtranslation Deletions = 0
Deleted Maintained Objects = 0
Dangling Links = 0
Verification Failures = 0

NCL>

Regards,
Himanshu

[Volker Halle]

Himanshu,

if the 'CISCO' router is supposed to talk DECnet protocol to your OpenVMS machine, you should see it as a router adjaceny on your local LAN:

$ MC NCL SHOW ROUT CIRC csmacd-0 ADJ *

(assuming your DECnet routing circuit is called csmacd-0 - if not, find out with MC NCL SHOW ROUT CIRC *).

The router also needs to have a DECnet address, find out from your network people and try SET HOST address. It will probably fail, but - according to the error message - you can find out, if the node is reachable.

Volker.

[Robert_Boyd]

Himanshu,

are the 2 DECnet systems in the same DECnet area? The 2 systems and the router need to be in the same DECnet area at a minimum unless there is an area router between them.



Robert

[Colin Butcher]

The term "DECnet over IP" is used to describe how DECnet will use IP as a transport from the VMS machine. It's a feature of DECnet-Plus (or DECnet/OSI on pre V7.1 systems) where DECnet uses TCPIP (or UCX) as a pseudo-transport layer, communicating through the PWIP driver and making use of the RFC1006 and 1006-Plus mechanisms (ports 102 and 399).

What you describe here sounds as if you talk DECnet from the VMS systems, but the Cisco routers encapsulate the DECnet packets inside TCPIP packets and then reconstitute the DECnet traffic at the far end. So, each LAN will have what appears to be a DECnet router and the LAN will have both DECnet and TCPIP protocols active on it (along wih a load of other protocols too). That's not "DECnet over IP" in the VMS sense.

So, normal DECnet routing would apply - you need to have all the DECnet systems in the same DECnet area (I'm assuming that you're using Phase IV compatible addressing), unless you can configure the Cisco routers to do DECnet Phase IV Level 2 (area) routing, in which case you would have each LAN as a separate DECnet area.

If you've only a small number of DECnet machines then I'd probably set them up as being in the same DECnet area and live with a trivial amount of extra WAN traffic as a consequence.

You should see the router as an adjacency on your LAN connection. The command "NCL SHOW ROUT CIRC ADJ * ALL" should show you the router on your end system. Both machines should see their local routers at DECnet level. The routers should see each other over their WAN link carrying the encapsulated DECnet traffic.

An alternative is indeed to use "DECnet over IP" where your VMS systems talk TCIP and you use TCPIP only routing. That can save you money at the routers, but you need to understand the WAN configuration and the routing / redundant path network design because you're now entirely reliant on your TCPIP network design and won't be able to use the DECnet Phase V capabilities of load balancing ovr all available paths.

You might find these helpful to give some of the DECnet Phase IV, Phase IV and DECnet over IP background:

http://www.downloads.xdelta.co.uk/bootcamp%20jun2005/i220_decnet-plus_issue_1-1_10jun2004.pdf

http://h71000.www7.hp.com/openvms/journal/v5/index.html#decnet

Cheers, Colin.

[Colin Butcher]

One other thing - I didn't see if you'd got any ECOs in place for DECnet/OSI V6.3 - a LOT of things got fixed back then (I broke most of them while testing stuff a few years back) and there's been another ECO released recently for DECnet/OSI V6.3 - ECO17.

See here:

http://www4.itrc.hp.com/service/patch/patchDetail.do?BC=patch.breadcrumb.main|patch.breadcrumb.search|&patchid=VAX_DNVOSIECO17-V0603&context=openvms:vax:6.2

You probably want to check the patch status of the V6.2 system while you're at it.

[Michael Yu_3]

Hi Himanshu,

I assumed that you are using Phase IV compatible addresses for your systems running DECnet/OSI.

When you said you could not set host to the remote machine, were you using the node name or node address? If you were using node name, try set host x.y where x.y is the node address for the remote system.

If it is OK, then the problem is just in the DECnet CDI cache and you can use the following to flush the cache.

NCL> flush session control naming cache entry "*"

After that set host nodename should be OK.

Thanks and regards.

Michael