kube-dns keeps restarting with kubenetes on coreos

kube-dns has a readiness probe that tries resolving trough the Service IP of kube-dns. Is it possible that there is a problem with your Service network?

Check out the answer and solution here:kubernetes service IPs not reachable

Lots of great debugging info here, thanks!

This is the clincher:

# curl https://10.3.0.1 --insecureUnauthorized

You got an unauthorized response because you didn't pass a client cert, but that's fine, it's not what we're after. This proves that kube-proxy is working as expected and is accessible. Your rkt logs:

Mar 22 22:17:58 kube-dns-3675956729-sthcv kubedns[8]: E0322 22:17:58.619254       8 reflector.go:199] pkg/dns/dns.go:145: Failed to list *api.Endpoints: Get https://10.3.0.1:443/api/v1/endpoints?resourceVersion=0: dial tcp 10.3.0.1:443: connect: network is unreachable

Are indicating that the containers that there's network connectivity issues inside the containers, which indicates to me that you haven't configured container networking/CNI.

Please have a read through this document: https://coreos.com/rkt/docs/latest/networking/overview.html

You may also have to reconfigure calico, there's some more information that here: http://docs.projectcalico.org/master/getting-started/rkt/