[j-nsp] Solved - Does a L3VPN RR require routing-instance for each VRF?

[Phil Mayers]

On 29/11/11 14:37, Eric Van Tol wrote:

> Basic question - I'm assuming that you have "cluster x.x.x.x"
> configured under the BGP group or individual BGP neighbor?  This is
> the only reason I could see this occurring, is if you were missing
> the cluster statement, or it was being overridden by something else.

Bingo, thanks. That's it.

I did not have a "cluster" statement.

The reason is that, as yet, this new RR does not *have* any clients, so 
I had not created the "group" statement. The only BGP peers are the 
existing route-reflectors, which of course go in a different group. i.e. 
I only had:

protocols {
     bgp {
         group Core {
             type internal;
             local-address z.z.z.z;
             family inet {
                 unicast;
                 multicast;
             }
             family inet-vpn {
                 unicast;
             }
             family inet6 {
                 unicast;
             }
             family inet6-vpn {
                 unicast;
             }
             family inet-mvpn {
                 signaling;
             }
             family inet-mdt {
                 signaling;
             }
             peer-as 64580;
             neighbor x.x.x.x;
             neighbor y.y.y.y;
         }
     }
}

When I added the "group" and set a "cluster" (even though there are no 
neighbours in it...) the "bgp.l3vpn.0" routing table suddenly magically 
appeared.

Cheers,
Phil