cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
482
Views
3
Helpful
7
Replies

(probably simple) BGP Question

riedmueller
Level 1
Level 1

I have 6 2-router pods. Each router has a serial link to the other router in the pod and a serial link to a different pod. The intra-pod link is running EIGRP. The inter-pod link is running BGP.

Each pod has it's own AS # (e.g. pod1 is 65001, pod2 is 65002).

When BGP syncs up, I am able to see the router in the adjacent pod, but no further. If I add EIGRP redistribution, I am able to see the router in the adjacent pod and the EIGRP network in the adjacent pod, but no further.

How do I configure the pods to allow BGP routes from distant pods to transit across the network? Do I have to run iBGP inside the pod's network to allow this? If so...can I just dump EIGRP?

This is a lab environment, so no users are impacted..just trying to make it work

1 Accepted Solution

Accepted Solutions

cpubob
Level 1
Level 1

Please include a diagram. From the sounds of the problem and your setup (still a bit of a mystery) it would seem you have something that looks like my diagram (attached). If this is true, then yes, you need ibgp between the peers in the pod or you need to redistribute bgp into eigrp all over the place (go with the first option). If you peer your intra pod peers with ibgp and you use the directly connected interface address's, then you can remove eigrp. If you peer with the loopback addresses then you'd still need eigrp or static routes.

Don't forget to resolve the next-hop issue you will have with the ebgp routes that are advertised to the ibgp neighbor. Either put the ebgp neighbor link into bgp or eigrp or use the next-hop self command toward the ibgp neighbor.

View solution in original post

7 Replies 7

nordick26
Level 1
Level 1

Hi,

at the border router (pod router running EIGRP and BGP) you'd need to redistribute the routes from eigrp to bgp or to use network command under bgp about the eigrp routes.

*remark:

When you do the redistrubution under bgp at border router like:

redistribute eigrp xxx

do not forget, that this eigrp routes have to be in your routing tabel. For example, if this routes are directly connected, redistribute eigrp will not reditribute these routes into bgp.

In this case, you'll need to do:

redistribute connected

under bgp process.

If any problem, you can post your config or some show commands outputs.

HTH

Regards

Ivan

cpubob
Level 1
Level 1

Please include a diagram. From the sounds of the problem and your setup (still a bit of a mystery) it would seem you have something that looks like my diagram (attached). If this is true, then yes, you need ibgp between the peers in the pod or you need to redistribute bgp into eigrp all over the place (go with the first option). If you peer your intra pod peers with ibgp and you use the directly connected interface address's, then you can remove eigrp. If you peer with the loopback addresses then you'd still need eigrp or static routes.

Don't forget to resolve the next-hop issue you will have with the ebgp routes that are advertised to the ibgp neighbor. Either put the ebgp neighbor link into bgp or eigrp or use the next-hop self command toward the ibgp neighbor.

Bob,

You are right on with the diagram. Seems I have some more work to do.

Thanks!

Hi Bob,

I tried sendig a post before but it did not get through.So here I go agaon :)

I'm a newbie with BGP and I'm not very clear about the comment "If you peer with the loopback addresses then you'd still need eigrp or static routes" made by you.

Could we use update-source command for the loopbacks and run ibgp instead of eigrp?

cheers

Hi Bob,

I'm a newbie with BGP and I'm not very clear about the comment "If you peer with the loopback addresses then you'd still need eigrp or static routes" made by you.

Could we use update-source command for the loopbacks and run ibgp instead of eigrp?

cheers

It all depends on which ip address you use in your neighbor statement. Say you have R1 connected to R2. The segment between R1 and R2 is 10.0.12.x/24, R1 is 10.0.12.1 and R2 is 10.0.12.2. R1's loopback is 1.1.1.1/32. R2's loopback is 2.2.2.2/32.

On R1, if your bgp neighbor statement uses R2's interface ip, 10.0.12.2, then you wouldn't necessarily need any igp running because R1 can form a ibgp peer with R2 using the directly connected interface between them.

On the other hand, if R1 uses R2's loopback in its neighbor statement, 2.2.2.2, then R1 will need a route in its routing table to reach 2.2.2.2. This can either be a static route in R1 or from a igp like ospf or eigrp.

The need for the update-source comes in when you have both routers, R1 and R2, each peering with the loopback of the opposite router. So for R1, its going to attempt to connect to 2.2.2.2, and when it looks up 2.2.2.2 in the routing table, it knows that destination is reachable out the interface 10.0.12.1. Therefore it attempts to create a tcp connection from 10.0.12.1 to 2.2.2.2. But R2 is expecting to neighbor with 1.1.1.1. When R2 gets a connection from 10.0.12.1, it doesn't know about that peer so your session doesn't come up. To fix that, you need to use the update-source command on R1 and R2 so the routers source the connection from the ip that the neighbor is expecting.

Hope that clarifies things a little bit. It made sense when I typed it but that doesn't mean what i typed makes any sense. :-)

-Bob

Hi Bob,

It certainly cleared my doubts.Many thanks for the detailed explanation.It was very kind of you.

cheers,

JA

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: