Trying to understand problems that occur when redistributing between two OSPF processes

[g.mundell]

Hi all, I'm currently brushing up on my OSPF and trying to understand the problems that can occur when redistributing between two OSPF processes. I have read and understand (I think!) the issues caused by the fact that the same route submitted by two different OSPF processes may not necessarily follow the OSPF rules that one would expect - for example, OSPF preferring intra-area routes to inter-area routes to external routes, but only within the same process. So, if the same route is submitted from two different processes, that rule goes out the window.

But I'm having some difficulty getting my head around the idea of setting the administrative distance lower in one OSPF process to prefer one domain over the other. I just can't quite follow the example described in this document:

http://www.cisco.com/c/en/us/support/docs/ip/open-shortest-path-first-ospf/4170-ospfprocesses.html#twored

Specifically, in figure 4 where two external networks - external network "N" originating in OSPF domain 1, and external network "M" originating in OSPF domain 2 - are redistributed via two ASBRs. The explanation states:

This sequence of events could occur occur:

1. Router A (Router B) redistributes M into Domain 1, and external M will reach Router B (Router A).

2. Because the administrative distance of Domain 1 is lower than Domain 2, Router A (Router B) will install M through Domain 1 and will set to maxage its previous originated LSA (event 1) into Domain 1.

3. Because M has been set to maxage in Domain 2, Router A (Router B) will install M though Domain 2 and, therefore, will redistribute M into Domain 2.

4. Same as event 1.

I can't quite work my way through this. I guess it must have something to do with the redistribution of "M" from domain 2 into domain 1 being learned by both ASBRs due to the lower administrative distance assigned to external routes in domain 1, and the original routes through domain 2 being deleted, but then I can't follow the rest of the description. And I can't understand why this would be a problem for network "M" in OSPF domain 2, but NOT for network "N" in OSPF domain 1.

Any explanation gratefully received!

Thanks, Graham

[Vasilii Mikhailovskii]

Hello.

You are right - whenever A and B learns about "M" from Domain 2, they craft LSA for domain 1 and inject it simultaneously. They learn each other's LSAs simultaneously and withdraw (set timer to 3600) for previous LSAs. And it might flap infinitely.

If they don't learn LSA simultaneously (let's say that A is much faster then B), then there will be no flaps, but B would learn all Domain 2 routes (not just redistributed) via Domain 1.

And later you will observe routing loop (when you stop advertising M from D): A knows "M" from Domain 2 and injects into Domain 1, B knows from A via Domain 1 and injects into Domain 2... so "M" stays in the routing tables due to mutual redistribution.

You don't have similar (flap) issue with network "N", because admin distance is lower for Domain 1, so both routers would never prefer OSPF via Domain 2! But having no issue with route flaps, you still will observe routing loop if you stop advertising "N" from C.

[g.mundell]

Thank you for your answer Vasilii, I think I am starting to get my head around this now!