We have a Jenkins master pod in K8s and jnlp container is based of official Jenkins image jenkins/jenkins:lts which I believe is alpine based. This spins up a Jenkins slave pod for each build which has a jnlp container and a busybox one. We have a weird behavior around with DNS. 1. Jnlp container in Jenkins Master pod can resolve all the internal and external hosts. 2. Jnlp container within the Jenkins Slave pod is unable to resolve any external hosts like google.com, and internal hosts. We checked the resolv.conf in that container and the name server corresponds correctly with kube-dns nameserver 10.96.0.10 . Internal hosts are not resolved, but we can hit them with IP’s. 3. The busybox container in the Slave pod can meanwhile resolve all the external hosts, but cannot resolve any internal hosts

If we introduced a maven-alpine container within the Jenkins slave pod, the behavior is identical to jnlp within slave pod where it cannot resolve any internal or external hosts. Any pointers and help would be appreciated