BFD for VPNv6 Static Routes

Load bfd.vpnv6.init.cfg

#IOS-XE
config replace flash:bfd.vpnv6.init.cfg

Configure a static route on R2 to R1’s IPv6 loopback (2001::1/128). Ensure R8 can reach this. Run BFD between R1 and R2 so that R2 withdraws the IPv6 route if R1 goes down.

Note that the XR routers are not used in this lab.

Answer

#R1
int gi2.12
 bfd interval 750 min_rx 750 multiplier 3
!
ipv6 route static bfd gi2.12 2001:10:1:2::2
ipv6 route ::/0 gi2.12 2001:10:1:2::2

#R2
int gi2.12
 bfd interval 750 min_rx 750 multiplier 3
!
ipv6 route static bfd vrf VPN_A gi2.12 2001:10:1:2::1
ipv6 route vrf VPN_A 2001::1/128 gi2.12 2001:10:1:2::1
!
router bgp 100
 add ipv6 vrf VPN_A
  redistribute static

Explanation

Strangely, while IPv4 static routes in a VRF require multi-hop BFD sessions, IPv6 static routes in a VRF allow for single-hop BFD sessions. These are much easier to configure.

On R1, we configure normal static routing with BFD as usual. Remember that with single-hop BFD sessions, the static route must reference an outgoing interface.

On R2, we simply refer to the VRF with the static route commands. Remember to redistribute static into BGP VPNv6 so that R8 can reach R1’s loopback.

Verification

We can verify the BFD session using show bfd neighbors detail on R2. We can see that IPv6 does not allow the use of the echo function, even though this is a regular single-hop BFD session. Also, we don’t see any reference to the VRF this session is running inside (VPN_A).

If we bring down Gi2.12 on R1, we should see that R2 removes the static route from the RIB and withdraws the VPNv6 route. Notice below that the astrik is missing from the static route in show ipv6 static vrf detail:

We can see on R5 that the VPNv6 route was withdrawn: