UCX Show route

[dufoem]

We have a Frame connection and a Internet connection on two different pipes comming into our network. The frame goes through our Cisco router. Our IP on the Cisco is 10.0.0.4 which I have setup our UCX route on our Alpha to route for our remote site that are on the Frame. Our gateway on our firewall is 10.0.0.1..

Maybe this will explain what I've done on our Cisco router.

config t
interface ethernet 0
no ip address 207.107.145.195 255.255.255.224 secondary
no ip address 216.208.33.193 255.255.255.224
ip address 10.0.0.4 255.255.255.0
exit
exit
copy running-config startup-config

Remove the old BGS Internet Feed Sub-Interface:

config t
no interface serial 0.10
exit
copy running-config startup-config

Modify the Routing Protocol so remote sites Learn of the changes:

config t
router eigrp 1
no network 207.107.145.0
network 10.0.0.0
exit
copy running-config startup-config

Change the Router's Default Gateway to the new Internet Feed:

config t
no ip route 0.0.0.0 0.0.0.0 serial 0.10
ip route 0.0.0.0 0.0.0.0 10.0.0.1
exit
copy running-config startup-config

Here's what I've done on our Alpha UCX settings.
UCX>set route/gate=10.0.0.4/def/perm
UCX> set route 220.0.6.0 /gateway=10.0.0.4
UCX> show route
AN 220.0.6.0 10.0.0.4

UCX> show route /perm

PERMANENT database

Type Destination Gateway

PN 0.0.0.0 10.0.0.4
PN 220.0.6.0 10.0.0.4

Now whenever a remote site looses power, the Alpha UCX route sets the gateway to 10.0.0.1. I have to manually go into UCX and change the gateway to point to 10.0.0.4.

We're running OpenVMS 6.2-1H3, UCX V 4.1.
We have users using the Frame and others using Telent using their ISP to connect to our Alpha box.

Does anyone know why this is happening and how I can fix so I don't have to keep going in and make changes whenever there is a power outage?

TIA,

Mike

[Willem Grooters]

Mike,


Now whenever a remote site looses power


That means, ths CISCO can no longer access 'the other side'. result, like you stated:


Change the Router's Default Gateway to the new Internet Feed:

config t
no ip route 0.0.0.0 0.0.0.0 serial 0.10
ip route 0.0.0.0 0.0.0.0 10.0.0.1
exit
copy running-config startup-config


That will mean that the default route for the router is 10.0.0.1, and if the 'otehr side' is lost, all traffic will go that way.

UCX will take that address as well - as advertised by the router. At least, that is what I conclude, perhaps UCX optimizes: Default 10.0.0.4 = routed on that CISCO to 10.0.0.1 (since it lost the other side and will then use it's default). Quite likely it's the same network - so why bother that router - send it over to the other one directly.

It might be an idea (i'm not that familiar with it) that another routing advertisement protocol could prevent this.

HTH (FWIIW)

Willem

[Martin P.J. Zinser]

Hi Mike,

I've seen exactly the same behaviour with
recent versions of TCP/IP (aka UCX) too.
Since these routers are not under our control
in this specific case we ended up writing a little script to reset the routes if needed...

Greetings, Martin

[Ian Miller.]

Is the problem is that the vms system fails over to the alternate route when the primary route is unavailable and does not return to using the primary route when it becomes available?

Is this a bad thing? It may be deliberate to prevent problems when a route fails and recovers briefly and fails again.

[Martin P.J. Zinser]

Hi Ian,

at least for us the problem was that the router was lying ;-), i.e. it advertised the route obviously but we could not get to the target actually since we were blocked by a firewall after the router.

And yes, this is more of an organisatorial than a technical problem, but hey aren't many problems in IT of that kind?

Greetings, Martin