Why is k3s creating /tmp/k3s-resolve.conf with nameserver 8.8.8.8?

[the-scott-hand]

Context:
I am spinning up a cluster using k3d. Inside the container is a file /etc/resolv.conf. Also inside the contain is a file /tmp/k3s-resolv.conf with the contents nameserver: 8.8.8.8. I am running in an air gapped network and do not want to reach out to 8.8.8.8.

Why is /tmp/k3s-resolv.conf being created? How can I avoid it being created/used? From what I can tell if I have a valid /etc/resolv.conf this file should not be made as per https://github.com/k3s-io/k3s/blob/master/pkg/agent/config/config.go#L312-L328 (which maybe i am not fully understanding?).

[ChristianCiach]

I found this comment: #53 (comment)

The settings are configurable in that we will either take a --resolv-conf flag to pass down to the kubelet, or a K3S_RESOLV_CONF environment variable will work also. We now try to use system resolv.conf files (from /etc & systemd), and will create a /tmp/k3s-resolv.conf file with nameserver 8.8.8.8 if nameservers in the system files are not global unicast ips.

You can provide your own resolv.conf to K3s, but I don't know how to do that when using K3d.

[brandond]

If you have IPv6 resolvers in your resolv.conf, k3s will generate its own using 8.8.8.8, as using IPv6 upstream DNS servers will break DNS in pods unless you have dual-stack networking. You might go ask in the K3d repo what's the best way to use your own resolv.conf.

[the-scott-hand]

That makes sense, I am beginning to suspect that the default provided /etc/resolv.conf in the container is considered invalid by the code i posted. Thanks!