I have Consul cluster on IPv6 assembled by Ansible. I added a new agent with Ansible, unfortunately I set IPv4 address as address for Consul. Now I can’t remove this bad agent. I tried force-leave (on consul server instances) with no luck. The member disappeared, but agent still trying to ping the member - filling log with Failed to send ping: write udp … sendto: network is unreachable (because consul trying ping IPv4 from IPv6). Member with bad IP is stopped for more than 72 hours.
Is something what I can do?
Did you manage to get this working? There is an issue raised in GitHub for the same Can't delete ipv4 node from ipv6 consul cluster · Issue #7691 · hashicorp/consul · GitHub.
At least from my testing what I have observed is, you can get rid of this node completely if you can bring down (stop) the whole cluster (including
clients) and bring them back again. I know this is not an ideal option especially for Production scenarios, but it should work.
From the logs, you will find that when you start the servers back, the stale node gets reaped properly.