The Vagrant set up assumes you are using kubectl from the same machine running Vagrant and references a bunch of generated TLS assets. You can see which in the provided kubectl config file:
https://github.com/coreos/coreos-kubernetes/blob/master/multi-node/vagrant/kubeconfig <https://github.com/coreos/coreos-kubernetes/blob/master/multi-node/vagrant/kubeconfig> You should be able to take those assets and SCP them onto one of the CoreOS nodes, along with the kubectl config. You maybe need to change that IP address to your public IP. Run kubectl similar to: kubectl —kubeconfig=“/path/to/config" More advanced, you should be able to cross-compile the kubectl Go code into a Windows binary. It might be worth opening an issue on the Kubernetes repo asking for them to start Windows releases or possibly use one of these community builds: https://github.com/eirslett/kubectl-windows <https://github.com/eirslett/kubectl-windows> If you did that and it still doesn’t show any pods, but you get no errors, be sure you are looking at all namespaces: kubectl get pods --all-namespaces You should see an API server, a controller manager, proxies, etc. - Rob > On Jun 23, 2016, at 7:28 PM, tech qi3 <[email protected]> wrote: > > > Hi, i followed the guideline in coreos website to understand the setup of > kubernetes coreos multi-node with ssl clusters. It seems everything work fine > except when I used kubectl get pods and return nothing. Aren't it suppose to > show the kubernetes component's pods? since all the components were created > in pod inside the master node. But when i check with docker ps, all > containers is up and running and calico is running with rkt. Does this have > something to do with the wrapper sript or am i missing something here? > > By the way I setup all the nodes with vagrant. I install kubectl within the > master node instead of ssh from elsewhere since i am using windows.
