Hey there! You've asked a great question about a very important concept in Kubernetes.
Marking a node as 'unschedulable' (which is what kubectl cordon does) means that Kubernetes will stop scheduling new pods onto that node.
Here's why this is useful:
- Maintenance: If you need to perform maintenance on a node (like upgrading the operating system, patching security vulnerabilities, or replacing hardware), you wouldn't want new applications to start running on it. Cordoning the node ensures no new pods land there, allowing you to safely drain the existing pods (using
kubectl drain) and then perform your maintenance. - Troubleshooting: If a node is experiencing issues, you might want to temporarily stop new workloads from being placed on it while you investigate and resolve the problem.
- Deprovisioning: When you plan to remove a node from your cluster, you first cordon it to ensure no new pods are scheduled there before gracefully terminating the existing ones.
It's important to differentiate this from a node being entirely offline or unhealthy. An unschedulable node can still run existing pods, but it won't accept any new ones. When the maintenance or issue is resolved, you can then mark the node as 'schedulable' again using kubectl uncordon.
Keep up the great work learning Kubernetes! Let me know if you have more questions as you go through the lab.
