I have a query regarding implementation of cluster-ids for route-reflectors. Say, I have 5 routers R1,R2,R3,R4 and R5 with R1 and R2 acting as a route-reflectors for the IPV4 address-family. I configure the cluster-id as 1.1.1.1 in both R1 and R2 forming a single cluster.
Now, in the same setup I enable IPv6, but only R2 is configured as the route-reflector. In such a scenario, when I check for the IPv6 routes in R1, only the R2 lopback IPv6 address is seen in the BGP tables. Other loopback addresses are dropped. I enabled the BGP updates debugs in R1 and could see the updates for other loopback addresses being denied. The reason being the updates are reflected and the cluster-id has its own cluster-id and hence dropped.
So does this mean, the route-reflectors with same cluster-id, should always have the same peers for all the address families or am I doing wrong in the configuration?
thanks
sbhas