Search code examples
kubernetescorednscalico

How do I debug this Kubernetes coreDNS error?

What does this error from my coredns pod log mean and how do I debug it?

[ERROR] plugin/errors: 2 2858211404501823821.6843583298703021155. HINFO: read udp 192.168.27.16:47449->67.207.67.3:53: i/o timeout

The behavior is odd.

A single test pod will execute a curl command correctly, but the network will not. Also each node is able to speak with each of the other nodes.

To my knowledge I have not changed any relevant configurations since the network last functioned "as expected."

UPDATE: So I do not know if this counts as a solution, but I deleted all pods (including coreDNS) and allowed them to restart. The system now works.

I will keep this question up and mark as solved just in case anyone does not know this nifty command (do not use on a production cluster)

kubectl delete po -A --all
Solution

  • I deleted all pods (including coreDNS) and allowed them to restart. The system now works.

    I will keep this question up and mark as solved just in case anyone does not know this nifty command (do not use on a production cluster)

    kubectl delete po -A --all

    Another way to do this (probably safer) is:

    kubectl -n kube-system rollout restart deployment coredns

    Thanks to @Richard_Bateman