Kubernetes has its own internal DNS service to provide service discovery between pods and services inside of the cluster. You can find more information about this at DNS for Services and Pods.
In MetaKube clusters we provide a highly available installation of CoreNDS as a cluster internal DNS solution.
Due to the way service discovery in Kubernetes works, the internal DNS resolution is a critical part of the cluster. To increase DNS performance and prevent DNS timeouts within the cluster we also provide a node-local-dns cache pod on every cluster node. These pods also run CoreDNS as the DNS cache. It works by running with hostNetwork:true and creating a dedicated dummy interface with a link local IP (169.254.20.10/32 by default) to listen for DNS queries. The cache instances connect to the cluster wide DNS service in case of cache misses.
For more information also have a look at the Nodelocal DNS Cache GitHub Page.
MetaKube allows the user to add additional configurations to CoreDNS and NodeLocal DNS Cache. This is achieved by creating a ConfigMap as follows:
Create a ConfigMap with the name dns-extra-configs in the Namespace kube-system, containing a valid CoreDNS style configuration file.
The configuration file name must be Corefile or end in Corefile as shown below.
cat <<'EOF' | kubectl apply --namespace kube-system -f -
apiVersion: v1
data:
Corefile: |
<Add zone information> {
<Add additional DNS configurations here>
}
kind: ConfigMap
metadata:
name: dns-extra-configs
namespace: kube-system
EOFOnce the ConfigMap has been created, use the following code block to restart all CoreDNS and NodeLocal DNS Cache pods to load the newly added configuration:
kubectl rollout restart deployment coredns --namespace kube-system
kubectl rollout restart daemonset node-local-dns --namespace kube-system