Kubernetes

Nodes Commands

Manage the worker machines in your cluster. Learn to inspect node health, cordon for maintenance, drain workloads, and troubleshoot node-level issues.

11 commands

Pro Tips

Use 'kubectl top nodes' to see CPU and memory usage across all nodes (requires metrics-server).

Before maintenance: 'kubectl cordon node' prevents new pods, 'kubectl drain node' evicts existing ones.

Use node labels and nodeSelector/affinity to control which nodes run specific workloads.

Common Mistakes

Don't forget to 'kubectl uncordon' nodes after maintenance - they won't receive new pods otherwise.

Draining nodes with pods using local storage requires '--delete-emptydir-data' flag.

Commands

List nodes

$ kubectl get nodes

List all cluster nodes.

Get nodes with details

$ kubectl get nodes -o wide

List nodes with IP and version info.

Describe node

$ kubectl describe node <node-name>

Show detailed node information.

Node resource usage

$ kubectl top nodes

Show CPU and memory usage per node.

Cordon node

$ kubectl cordon <node-name>

Mark node as unschedulable.

Uncordon node

$ kubectl uncordon <node-name>

Mark node as schedulable again.

Drain node

$ kubectl drain <node> --ignore-daemonsets --delete-emptydir-data

Safely evict pods before maintenance.

Label node

$ kubectl label nodes <node-name> key=value

Add label to a node.

Taint node

$ kubectl taint nodes <node-name> key=value:NoSchedule

Add taint to node to repel pods.

Remove taint from node

$ kubectl taint nodes <node-name> key:NoSchedule-

Remove taint from a node.

Get worker nodes only

$ kubectl get nodes --selector='!node-role.kubernetes.io/control-plane'

List only worker nodes (exclude control plane).