talosctl uses public IP, even though internal IP should be used

[Stupremee]

Hello,

I have successfully created a Kubernetes Cluster on Hetzner using Talos.
Now I encountered a bug, where talosctl health suddenly doesn't work, because it tries to use the public IP for one of the nodes, which is unreachable.

The weird thing is, that for all other nodes, it uses the correct IP Addresses, as you can see from the first line of the health command output:

discovered nodes: ["10.0.65.1" "10.0.65.129" "10.0.64.1" "10.0.64.21" "<public IP of node>"]

I then executed to do talosctl get nodeaddresses with the endpoint being the node whose public IP is used.
And the weird thing is that for this node the public IP is the first element. So I guess that's the reason it choses the public IP.

For all other nodes, the first element in the list of addresses is the private IP.

What are some things I can try to resolve this issue?

[smira]

Please see #10848 (comment)