[Xorp-users] Route Redistribution issue in BGP - Kindly help asap.

[Atanu Ghosh]

Hi,

It should be possible to do what you want in a single term:
policy {
    policy-statement "routes_as1_to_as2" {
        term "reject_ospf4_to_ibgp_peer" {
            from {
                    protocol: "ospf4"
            }
            to {
                    neighbor: 10.5.11.1
            }
            then {
                    reject /*Rej ospf routes to IBGP Peer*/
            }
        }
    }
}

However it looks as if you have hit a bug with the neighbor statement:
http://www.xorp.org/bugzilla/show_bug.cgi?id=610

We will look into this and get back to you.

   Atanu.

>>>>> "Santhosh" == Santhosh Sundararaman <santhosh at ku.edu> writes:

    Santhosh> Hi,
    Santhosh> I have a BGP router that is peered to one E-BGP  peer and one I-BGP 
    Santhosh> peer. There are several routers in the local AS and OSPF is the IGP. I 
    Santhosh> am trying to redistribute the OSPF routes into BGP such that the OSPF 
    Santhosh> routes are advertised to the EBGP peer but not the IBGP peer.  The 
    Santhosh> following is the configuration I am using.

    Santhosh> protocols {
    Santhosh> bgp {
    Santhosh> bgp-id: 172.16.10.3
    Santhosh> local-as: 65001

    Santhosh> export: "routes_as1_to_as2"

    Santhosh> peer 172.16.10.1 { /* EBGP Peer*/
    Santhosh> local-ip: 172.16.10.3
    Santhosh> as: 65002
    Santhosh> next-hop: 172.16.10.3
    Santhosh> holdtime: 120
    Santhosh> ipv4-unicast: true
    Santhosh> }
       
    Santhosh> peer 10.5.11.1 { /*IBGP Peer*/
    Santhosh> local-ip: 10.10.11.2
    Santhosh> as: 65001
    Santhosh> next-hop: 10.10.11.2
    Santhosh> holdtime: 120
    Santhosh> ipv4-unicast: true
    Santhosh> }
    Santhosh> }   
    Santhosh> }

    Santhosh> policy {
    Santhosh> policy-statement "routes_as1_to_as2" {
    Santhosh> term "ospf_routes" {
    Santhosh> from {
    Santhosh> protocol: "ospf4"
    Santhosh> }
    Santhosh> to {
    Santhosh> neighbor: 172.16.10.1
    Santhosh> }
    Santhosh> then {
    Santhosh> accept /*adv ospf routes to EBGP Peer*/
    Santhosh> }
    Santhosh> }

    Santhosh> term "reject_ospf4_to_ibgp_peer" {
    Santhosh> from {
    Santhosh> protocol: "ospf4"
    Santhosh> }
    Santhosh> to {
    Santhosh> neighbor: 10.5.11.1
    Santhosh> }
    Santhosh> then {
    Santhosh> reject /*Rej ospf routes to IBGP Peer*/
    Santhosh> }
    Santhosh> }
    Santhosh> }
    Santhosh> }

    Santhosh> On using this policy the routes get advertised to both EBGP and IBGP 
    Santhosh> peers and "reject_ospf_to_ibgp_peer" term appears to be ignored. I have 
    Santhosh> tried specifying "reject_ospf_to_ibgp_peer" without any 
    Santhosh> to{neighbor:10.5.11.1}  rule, in which case the routes were not 
    Santhosh> advertised to any of the peers. Am I missing something??

    Santhosh> The address matched against the neighbor variable in the to {} rules, is 
    Santhosh> the address of the interface of the peer to which peering is establish 
    Santhosh> and not the bgp-id of the peer, is that correct or should it have been 
    Santhosh> the peers bgp-id instead.

    Santhosh> Also inside the bgp protocol construct when specifying the peers, should 
    Santhosh> the peer address be the bgp-id of the peer, or can it be any one of the 
    Santhosh> several interface addresses of the peer which may not be the bgp-id.

    Santhosh> Any help would be greatly appreciated.

    Santhosh> Thanks
    Santhosh> Santhosh

    Santhosh> _______________________________________________
    Santhosh> Xorp-users mailing list
    Santhosh> Xorp-users at xorp.org
    Santhosh> http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users