Static IPv6 routing with link-local next hops and a floating backup

Tasks

1. Enable IPv6 routing on both routers and address the LANs.
2. Give both inter-router links fixed link-local addresses and no global addresses.
3. On each router, add a static route to the other LAN via the primary link, using the link-local next hop.
4. Add a floating static route via the backup link with administrative distance 200.
5. Verify the routes, test reachability, then fail the primary link and check the backup takes over.

Solution

1. 1R1 interfacesR1

   ipv6 unicast-routing
   !
   interface GigabitEthernet0/0/0
    ipv6 address 2001:db8:acad:10::1/64
    no shutdown
   !
   interface GigabitEthernet0/0/1
    description Primary to R2
    ipv6 address fe80::1 link-local
    no shutdown
   !
   interface GigabitEthernet0/0/2
    description Backup to R2
    ipv6 address fe80::1 link-local
    no shutdown

   The same link-local address on two interfaces is legal, because each one is unique only on its own link. Configuring a link-local address enables IPv6 on the interface, so no global address is needed.

2. 2R1 static routesR1

   ipv6 route 2001:db8:acad:20::/64 GigabitEthernet0/0/1 fe80::2
   ipv6 route 2001:db8:acad:20::/64 GigabitEthernet0/0/2 fe80::2 200

   Both routes name fe80::2. Only the exit interface tells them apart, which is exactly why IOS requires it.

3. 3R2 interfaces and routesR2

   ipv6 unicast-routing
   !
   interface GigabitEthernet0/0/0
    ipv6 address 2001:db8:acad:20::1/64
    no shutdown
   !
   interface GigabitEthernet0/0/1
    description Primary to R1
    ipv6 address fe80::2 link-local
    no shutdown
   !
   interface GigabitEthernet0/0/2
    description Backup to R1
    ipv6 address fe80::2 link-local
    no shutdown
   !
   ipv6 route 2001:db8:acad:10::/64 GigabitEthernet0/0/1 fe80::1
   ipv6 route 2001:db8:acad:10::/64 GigabitEthernet0/0/2 fe80::1 200

Verify

R1# show ipv6 route static | begin ^S
S   2001:DB8:ACAD:20::/64 [1/0]
     via FE80::2, GigabitEthernet0/0/1

R1# show ipv6 neighbors GigabitEthernet0/0/1
IPv6 Address                              Age Link-layer Addr State Interface
FE80::2                                     0 a0b4.3c22.1401  REACH Gi0/0/1

R1# ping 2001:db8:acad:20::1 source GigabitEthernet0/0/0
Type escape sequence to abort.
Sending 5, 100-byte ICMP Echos to 2001:DB8:ACAD:20::1, timeout is 2 seconds:
Packet sent with a source address of 2001:DB8:ACAD:10::1
!!!!!
Success rate is 100 percent (5/5), round-trip min/avg/max = 1/1/1 ms

R1# show ipv6 route static | begin ^S
S   2001:DB8:ACAD:20::/64 [200/0]
     via FE80::2, GigabitEthernet0/0/2

Mistakes people make

• Writing ipv6 route 2001:db8:acad:20::/64 fe80::2 without the interface. IOS rejects it, since every link has fe80::/10.
• Using an interface-only route on Ethernet (ipv6 route 2001:db8:acad:20::/64 GigabitEthernet0/0/1). R1 then tries neighbor discovery for every destination and R2 does not answer for them.
• Adding the backup only on R1. Return traffic still needs a path, so mirror the routes on R2.
• Expecting the floating route to take over when R2 dies but the link stays up. A static route only follows interface state; add BFD or IP SLA tracking for that case.
• Forgetting that traceroute through these links shows no global address for the transit hops. ICMPv6 errors are sourced from another interface's global address.

On other platforms

set routing-options rib inet6.0 static route 2001:db8:acad:20::/64 qualified-next-hop fe80::2 interface ge-0/0/1.0
set routing-options rib inet6.0 static route 2001:db8:acad:20::/64 qualified-next-hop fe80::2 interface ge-0/0/2.0 preference 200

ip -6 route add 2001:db8:acad:20::/64 via fe80::2 dev eth1 metric 100
ip -6 route add 2001:db8:acad:20::/64 via fe80::2 dev eth2 metric 200