From: Kevin Baumgartner (kbaumgar@xxxxxxxxx)
Date: Wed Jan 24 2001 - 02:15:24 GMT-3
I'm surprised that the requirement is to use sync with route reflectors.
The recommended design is to do no sync. The reason being the igp routing
protocol (ospf) is not going to be in sync with bgp in this configuration.
There is no connectivity between R3 and R6 as far as BGP is concerned.
>From R6 to R3, BGP considers R4 the next hop router. But for OSPF it
most likely considers R3 as directly connected. The routing paths that
both OSPF and BGP use need to be the same for sync to be used.
One way you could get this to work is change the ospf cost on the link
betweem R4 and R6 lower such that for OSPF it's the best path to get to R3.
And the next hop router is R4. Then OSPF and BGP will be in sync.
Not the recommend way to do this but should work.
Try doing a extended traceroute from R6 to R3. My guess is that you
will see a routing loop between R6 and R4 and it never gets to R3.
Kevin
>
> All-
>
> I came across an scenerio that requires BGP to be sync.. Here is the simple
> topology. The IGP protocol is OSPF and the BGP AS is 600.
>
> R4-------------------------R6
> |
> |
> |
> |
> |
> |
> R3
>
> R4 is the route reflecter for R6 and R3. Each router has a loopback
> interface that I have to advertise via BGP. Since the lab requires all
> routers to be "synch" I redistribute the BGP loopback interfaces on all
> routers into OSPF. All interfaces show up via OSPF. R4 is synch with both
> the loopback interfaces on R6 and R3, but if I go to R6, R6 is synch with
> R4's loopback interface, but not with R3's loopback interface, and the same
> on R3. Even though R3's loopback interface shows up via OSPF and can ping
> R3's loopback interface from R6..
>
> Thoughts?
> Robert
>
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:27:42 GMT-3