Hagbarddenstore

Static DNS with DHCP on CoreOS

Published on 2015-06-24 10:39

Categories: coreos, infrastructure

Tags: coreos, dns, infrastructure

TL;DR;

How to set static DNS servers while still getting IP and routes from the DHCP server in CoreOS.

Problem

I’ve setup two servers running CoreOS and SkyDNS to provide DNS resolution in my AWS VPC, creating a DHCP option set which used the two CoreOS instances as the DNS servers, which meant that the CoreOS instances got themselves as their DNS server. This meant that whenever the instances where unavailable they couldn’t resolve any names, which meant I couldn’t pull Docker images.

The solution

To fix this problem, I figured I wanted to set a static DNS, which is easy with the following commands:

# echo "nameserver 8.8.8.8" > /etc/resolv.conf
# echo "nameserver 8.8.4.4" >> /etc/resolv.conf

But this solution doesn’t persist between boots or whenever the network service is restarted, since the dhcpclient will overwrite the /etc/resolv.conf file. So, to have a permament solution I needed to add the configuration into the cloud-config file. My initial cloud-config looked like this:

#cloud-config

coreos:
  etcd:
    discovery: https://discovery.etcd.io/2d8e9b50bc44d85d0e002767a86eff3a
    addr: $private_ipv4:4001
    peer-addr: $private_ipv4:7001
  units:
    - name: etcd.service
      command: start
    - name: fleet.service
      command: start
    - name: 00-eth0.network
      runtime: true
      content: |
        [Match]
        Name=eth0

        [Network]
        DHCP=yes
        DNS=8.8.8.8
        DNS=8.8.4.4

Which looked like it should work and it did work, but with the funny exception that it used the DNS servers I had specified and also the one from DHCP. So my /etc/resolv.conf looked something like this:

nameserver 8.8.8.8
nameserver 10.0.0.10
nameserver 10.0.0.20
nameserver 8.8.4.4

So the /etc/resolv.conf now contains two correct lines and two wrong lines. The 10.0.0.{10,20} addresses are the addresses to the CoreOS instances.

So, after looking through the systemd.network documentation some more, I found that you should add the following lines to your cloud-config:

[DHCP]
UseDNS=false

This gave me the following final cloud-config:

#cloud-config

coreos:
  etcd:
    discovery: https://discovery.etcd.io/2d8e9b50bc44d85d0e002767a86eff3a
    addr: $private_ipv4:4001
    peer-addr: $private_ipv4:7001
  units:
    - name: etcd.service
      command: start
    - name: fleet.service
      command: start
    - name: 00-eth0.network
      runtime: true
      content: |
        [Match]
        Name=eth0

        [Network]
        DHCP=yes
        DNS=8.8.8.8
        DNS=8.8.4.4

        [DHCP]
        UseDNS=false

Which gave me the following /etc/resolv.conf file:

nameserver 8.8.8.8
nameserver 8.8.4.4

Success! Now I have a CoreOS cluster with static DNS servers which persist between reboots and network restarts.

Warning

Do not copy and paste the cloud-config examples into your cloud-config before you edit the discovery token.