I have installed a kubernetes cluster manually according to this guide: https://kubernetes.io/docs/getting-started-guides/centos/centos_manual_config/
I have installed the kubedns and kubernetes-dashboard addons, but when trying to spin up a heapster deployment, the dashboard warns me that "kubelet does not have ClusterDNS IP configured". However, I believe this to be incorrect. In the kubelet config of my node I have specified:
KUBELET_ARGS="--cluster-domain=cluster.local --cluster-dns=10.250.100.53"
The IP being the flannel IP of my kubeDNS service.
I believe this is working, since when I let a pod do an nslookup without specifying the server, it works.
[root@myKubeMaster]# kubectl exec busybox -- nslookup kubernetes.default.svc.cluster.local
Server: 10.250.100.53
Address 1: 10.250.100.53 kube-dns.kube-system.svc.cluster.local
Name: kubernetes.default.svc.cluster.local
Address 1: 10.250.100.1 kubernetes.default.svc.cluster.local
Where does this error come from?
Seems like there could be an error in how the arguments get passed to the kubelet, could you try splitting out the arguments?(https://github.com/kubernetes/kubernetes/issues/27722#issuecomment-227893612)