Windows: Seeing intermittent DNS issues on k8s

General Discussions Windows
, ,
1

When operating in environments with high pod churn rates—where pods are frequently created and terminated—and multiple services are configured to reference a single deployment, we observed that network connectivity on some of the pods is failing on windows nodes.

On the pods where we have network issue, tried to do DNS lookup by exec into the pods and it failed to resolve the DNS.

I was going through the below PR and want to check whether any regression caused the network connectivity issue.
PR1: #135146
PR2: #133953

Our customers reported that they started seeing this issue from long time since k8s 1.31 and higher.

2

Thanks for reporting this.

Could you share a bit more detail about your setup (Windows version, container runtime, CNI, and kube-proxy mode)? Also, when DNS fails inside the pod, does nslookup to the CoreDNS service fail consistently or only intermittently?

It would also help to check kube-proxy and HNS logs on the affected Windows nodes to see if there are any networking errors when the issue occurs.