Docker containers fail to restart with ipv6 enabled with LLA fixed cidr

General
1

We are experimenting the docker development with ipv6 with Linked local address, but error occured when we restart the container.

Step to reproduce the error:

  1. As link local communication is needed, I overrided the /etc/docker/daemon.json file
{
  "ipv6": true,
  "fixed-cidr-v6": "fe80::1/64"
}
  1. Restart the docker service to apply the change
service docker restart
  1. Launch any docker container and kill it.
docker container run --detach nginx:latest
docker kill <container_id>
  1. Re-run the docker container
docker container run --detach nginx:latest

And it shows docker: Error response from daemon: Address already in use.

If we change the fixed-cidr-v6 field with another subnet range “2001:db8:1::/64”, the restart will work, however we cannot do the local communication that we want for experiment.

ip 6 route

::1 dev lo proto kernel metric 256 pref medium
fe80::/64 dev docker0 metric 1024 linkdown pref medium

docker version: 23.0.5

2

Are you aware that the prefix 2001:db8::/32 is used for documentation purposes only? That’s why you find it as placeholder in documentations.

I am not sure if fe80::1/64 is literarily used as ip in the subnet fe80::/64, or results in the subnet fe80::/64 being used. I guess you wanted to use fe80:1/64?

Personally, I would use an ULA instead of a LLA. Both will not allow internet access.