[c-nsp] Route Redistribution Problem (static route into OSPF into MBGP)

[Sam Stickland]

Hi,

I've got a strange route distribution issue. The router setup is this:

Cust Network 1
|           |
|           |
R1 ------- R2
|           |
|           |
R3 ------- R4
|           |
|           |
Cust Network 2

MPLS links are in use between all four routers, and all of them are in a
full MBGP mesh. This is a pretty standard MPLS VPN setup with the customer
running OSPF as their IGP.

The routers R1 and R2, (and R3 and R4) run OSPF with the customer network,
inside of a VRF "vrf-one". The customer routes are transported via MBGP over
the MPLS network (all inside "vrf-one"). The interfaces into the customer
network are in VRF vrf-one.

On R1 and R2 static routes (to a management network) are redistributed into
the customers OSPF (130).

The problem is that this route is getting distributed into Cust Network 1's
OSPF, but the route from there isn't getting redistributed into BGP, so that
Cust Network 2 never learns the route.

A sample configuration from R1 is:

ip vrf vrf-one
 rd 64513:1
 route-target export 64513:1
 route-target import 64513:1

router ospf 130 vrf vrf-one
 router-id 172.16.25.253
 log-adjacency-changes
 redistribute static metric 50 subnets
 redistribute bgp 64513 subnets
 network 172.16.25.128 0.0.0.3 area 0
 network 172.16.25.253 0.0.0.0 area 0

router bgp 64513
 no synchronization
 bgp router-id 172.16.26.254
 bgp cluster-id 167838050
 bgp log-neighbor-changes
 neighbor 172.16.26.252 remote-as 64513
 neighbor 172.16.26.252 update-source Loopback100
 neighbor 172.16.26.253 remote-as 64513
 neighbor 172.16.26.253 update-source Loopback100
 neighbor 172.16.26.255 remote-as 64513
 neighbor 172.16.26.255 update-source Loopback100
 no auto-summary

 address-family ipv4 vrf vrf-one
 redistribute ospf 130 vrf vrf-one match internal external 1 external 2
 no auto-summary
 no synchronization
 exit-address-family

ip route 10.200.20.0 255.255.255.0 172.16.27.129
ip route vrf vrf-one 10.200.20.0 255.255.255.0 172.16.27.129 global

The prefix 10.200.20.0/24 is the management network. Logging into one of the
routers on "Cust Network 1" shows that the route has been distributed into
the customers OSPF process (130).

However, from there the route doesn't make it into BGP and across the
network.

R1#sh ip bgp vpnv4 all nei 172.16.26.253 advertised-routes | inc 10.200.20

R1#sh ip route vrf vrf-one 10.200.20.0
Routing entry for 10.200.20.0/24
  Known via "static", distance 1, metric 0
  Redistributing via ospf 130
  Advertised by ospf 130 metric 50 subnets
  Routing Descriptor Blocks:
  * 172.16.27.129 (Default-IP-Routing-Table)
      Route metric is 0, traffic share count is 1

Why doesn't this route end up inside the BGP table?

Sam