- Configure
kube-scheduler
so that Pods are automatically assigned to a node - The Pod definition for this step:
- Doesn't set the node name to which the Pod should be deployed.
- Tolerates the
node.kubernetes.io/not-ready:NoSchedule
taint. This is needed because Kubernetes seems to taint the nodes while the control plane is not fully setup.
- The ServiceAccount admission controller is still disabled, as in the previous step.
- Start the VM and wait for the configuration. Then SSH into it.
vagrant up vagrant ssh
- Check that
etcd
,kube-apiserver
andkube-scheduler
are running as Docker containers.docker ps
- Create the test pod using
kubectl
and check that it is running. It may take a while for containers to be created due to image downloads.- This means that
kube-scheduler
picked up the new pod and assigned a node to it.
kubectl create -f pod.yaml kubectl get pods docker ps
- This means that
- Check that the pod's recent events include its scheduling.
kubectl describe pod nginx | grep -A5 Events
- Close the SSH connection and destroy the VM
logout vagrant destroy -f