Proxies and dual Nics

[matthew robey]

VMS 7.3-2 Decnet V over TCPIP V5.4 eco 4.

Each VMS system has dual nics configured with
seperate ip addresses and IP failover.
(Both addresses in DNS database (windows), and DNS ip address in local host database.)

When adding proxy NODE::USERNAME the first $DIR NODE:: works, the second fails.

ie
$ DIR NODE(1.1.1.1)::USER works
$ DIR NODE(1.1.1.2)::USER fails

It seems as if the remote node name translation to ip address stops at the first entry in the DNS database, instead of going through and reading the second ip address for the remote node.

I have worked around this by putting both addresses of the remote node in the local host database, but surely this is a bug ?

Matt

[Michael Yu_3]

Hi Matt,

If you are using IP failsafe, the outgoing connections are balanced across the available controllers using round robin.

When you use DECnet over IP, the source address in the Connect Request packet will contain the source IP address. Everytime you make an outgoing connect request, the source IP address will be different because of the IP failsafe.

When the call comes into the destination node, it is recognised as coming in from the remote IP address. The backtranslation from IP address to name is one-to-one which is different from the name to address translation which can be one-to-many. Hence multiple addresses are needed in your proxy database.

Thanks and regards.

Michael

[Colin Butcher]

This is one of the main areas where DECnet and IP differ in design terms. Think of DECnet addressing as "per node" and of IP addressing as "per interface".

A DECnet node name will correspond to many interfaces. An IP host name will correspond to a single IP address.

So, you need multiple names in the IP DNS to translate / back translate the addresses.

DECnet over IP uses IP all the way from the naming layer, through the PWIP driver and then uses IP as a pseudo-transport via the RFC1006 / 1006-Plus templates. No DECnet is involved, apart from the application layer programming interfaces.

If these systems are on the same pair of separate LANs (one interface per LAN) then I would strongly recommend that you use native DECnet Phase V between them, probably with Phase IV compatible addressing. It will give you dual-rail load balancing and automatic handling of failover. IP V4 does not do that.

failsafe IP is all about having an IP address for a service (not a node) which can move around all interfaces in a cluster (between interfaces on the same node first, then another node if the primary node for that PI address fails). It works best when you have a set of IP addresses, one for each service that your systems offer, and you can then allocate the interface order for each service address by setting up failsafe IP interfaces in the cluster in the order you choose.

Change the way you think about IP addresses - they are no longer per interface, but are per service and they can move around the available interfaces within the cluster.

This is a very useful description:

http://h71000.www7.hp.com/openvms/journal/v2/articles/tcpip.pdf

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

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

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

Cheers, Colin.

[matthew robey]

Thanks for the replies.

The question still exists though as to why with two entries in DNS for the same node (different ip addresses), only the first one is being recognised for back translation. The second one is ignored, and hence fails, unless in the local host database.

Matt

[Michael Yu_3]

Hi Matt,

The host name can have several IP addresses registered with the DNS server. However for back-translation, each IP address refers to a unique host name. The DNS server uses different databases for lookup and reverse lookup.

Thanks and regards.

Michael