From: Peter van Oene (pvo@usermail.com)
Date: Mon Apr 12 2004 - 11:34:32 GMT-3
At 10:27 AM 4/12/2004, Richard Dumoulin wrote:
> >Right. So what is the forwarding address in each of these cases? Explicit
> >to .2 in the first, and 0.0.0.0 in the second I'd imagine?
>
>Yes
>
> >Also, what are the implications if .2 is reachable, but by another
> >protocol, say eigrp to the rest of a multi-area OSPF domain. Really some
> >interesting stuff comes into play here that all deal with how the
> >forwarding address is set,
>
>Implications if the forwarding address is set to .2 ? .10 avoids an extra
>hop.
Inter area routers won't use this LSA if that is the case. Also, if R2 is
reachable by say EIGRP, inter area OSPF routers will ignore the type 5.
>Implications if the forwarding address is 0.0.0.0 ? Extra hop + icmp
>redirect ? I would have to run some debugs to see if .1 is sending
>redirects. But by hearing you there seems to be more ?
>
>Also .2 belongs to the 137.20.20.0/24 network which is part of the ospf
>cloud. So I don't see how .2 can be reachable by another protocol.
Often the case during transitions. Cisco loves to build big flat EIGRP
networks as they are easy to setup (ie little post sales cost to Cisco) and
keep other vendors out. Many of us spend a lot of time fixing these
networks and OSPF migrations are frequent. It is here, where forwarding
address control can really bite you if you ignore it.
> >and further what the restrictions are for using
> >type 5's with various forwarding addresses (ie you must be able to resolve
> >the forwarding address via an OSPF inter area route).
>
>Hehe, here I have no clue, would you shed some light ? I guess you are
>seeing deeper than me ...
Just the point that type 5 forwarding addresses need to be resolved by OSPF
inter area routes or OSPF itself ignores them. Much like BGP next-hop
recursion, OSPF recurses 5's
>--Richard
This archive was generated by hypermail 2.1.4 : Mon May 03 2004 - 19:48:46 GMT-3