Open Infrastructure

router1:~# ip addr add 2a0a:e5c0:100::4242/128 dev lo

• bird:

filter router1sane {
    if(net ~ [ 2a0a:e5c0:100::/40+ ]) then {
        accept;
    }
    reject;
}

protocol direct  {
    interface "dummy0";

#     ipv4 {
#       import filter from_loopback;
#       export none;
#     };

     ipv6 {
       import filter from_loopback;
       export none;
     };
 }

interfaces:

auto dummy0
iface dummy0 inet manual
    up ip link add $IFACE type dummy
    up ip link set $IFACE up
    down ip link del $IFACE type dummy
    post-up ip addr add 2a0a:e5c0:100::4242/128 dev $IFACE

The commands in the previous comment are enough to enable the IP address globally. However, it is not enough for enabling it locally, as the NDP process does not work without the switches knowing how to access it.

Checking switches:

 B    2a0a:e5c0:2:5::4242/128 [200/0]
       via 2a0a:e5c0:1:8:714a:8de7:67ca:7284, Vlan8

The router is actually receiving the NDP query:

16:43:17.121249 IP6 2a0a:e5c0:2:5:70c2:d35d:93a0:e367 > ff02::1:ff00:4242: ICMP6, neighbor solicitation, who has 2a0a:e5c0:2:5::4242, length 32

However router1.place6 does not answer it (likely due to being on the wrong interface)

Then again, the switches can reach the ipv6 address:

[admin@switch5-place6 ~]$ ping6 2a0a:e5c0:2:5::4242    
PING 2a0a:e5c0:2:5::4242(2a0a:e5c0:2:5::4242) 56 data bytes
64 bytes from 2a0a:e5c0:2:5::4242: icmp_seq=1 ttl=64 time=0.379 ms
64 bytes from 2a0a:e5c0:2:5::4242: icmp_seq=2 ttl=64 time=0.250 ms
^C
--- 2a0a:e5c0:2:5::4242 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1000ms
rtt min/avg/max/mdev = 0.250/0.314/0.379/0.066 ms
[admin@switch5-place6 ~]$ 

• So the problem so far is that NDP is not reaching dummy0.
• If we setup the virtual IP address on the real interface (bond0.15), then likely DAD will prevent it from functioning, as it is assigned twice.