0

I am trying to resolve a loop error with CoreDNS as part of a 2-node Kubernetes cluster. The issue is thoroughly documented (below), but I am not strong on dns and networking so am having trouble working through it.

Core DNS doco: https://coredns.io/plugins/loop/

Kubernetes manifestation: https://github.com/kubernetes/kubeadm/issues/1162

Environment:

  • Home network with no DNS DHCP served through router.
  • CIDR: 10.10.0.0/32
  • Two Ubuntu 16.x nodes in the cluster (Master is also a Minion)
  • /etc/resolv.conf contains:

    nameserver 127.0.1.1

Question

In order to get this working, do I need to set up a nameserver? If so, what should I use? (I have been reading about BIND.) If so, can you please provide some good entry-level info for getting my head around what I need to do?

Thanks!

Bryon
  • 131
  • 2
  • 1
    Have you seen [this](https://stackoverflow.com/questions/53075796/coredns-pods-have-crashloopbackoff-or-error-state/53414041#53414041)? It looks like someone has an issue like yours with the DNS. – Manuel Florian Jan 14 '19 at 01:33

1 Answers1

1

I have worked out that yes you do need a DNS server for a multi-node cluster. The link @Manuel posted is essentially what I was doing and it needs to be done, but they infer you need an external DNS. The instructions for configuring the target DNS server are also incorrect; you cannot make the changes to /etc/resolv.conf the way it describes (or at least I couldn't).

Bryon
  • 131
  • 2
  • An update on changing the DNS in resolv.conf... See this topic where I posted my solution. https://serverfault.com/questions/230456/permanently-change-dns-settings-in-ubuntu-server/949063#949063 – Bryon Jan 15 '19 at 02:00