wireguard ipv6 between 2 (or more) sites

vu flag

I want to connect our office to servers with VMs in DC using wireguard and ipv6 ULA addresses.

The office ULA is FD58:1234:beef:cafe::/64. I want to use the same range for server + VMs in DC. Connection between both points is done using IPv4 public address of each end. Works great, tunnel is open, one site is FD58:1234:beef:cafe:fade::12/96 -router1 office- the other one FD58:1234:beef:cafe:fade::12/96 -router2 DC- I can ping each router from both themself as well as from machines in the office behind router1

On the DC server I setup ULA FD58:1234:beef:cafe:fade::7000/120 for VMs (KVM), interface being virbr0 with IP FD58:1234:beef:cafe:fade::7000/120. One VM has FD58:1234:beef:cafe:fade::7012/120 as IP. From this VM I can ping virbr0 as well as both routers IPs.

Problem is that when I try to ping a machine at office (eg behind router1) it take long time (between 100 to 500 seq) till I get an answer. Some moment during my tests I saw neighbor solicitation without answers so I installed ndp6 on each router. I played also with route metrics to give local ranges higher priority.

From tcpdump I see all the requests without answer and at a time answer come and it start to do the job. Both routers are Debian11.

If someone had a clue on this, would be great to share ;)



Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.