CKA Full Course 2024: Day 15/40 Kubernetes Node Affinity Explained

Here we go with another video and fresh labs we need to do. But before we get going, let’s discuss the differences between Node Affinity and Taints and Tolerations because I’ve got to keep it real, I was (more than) a bit confused.

The key difference between Node Affinity and Taints and Tolerations lies in how they control pod placement on nodes:

• Node Affinity: Specifies where pods can be scheduled based on node labels. It uses "required" or "preferred" criteria to influence pod placement.

• Taints and Tolerations: Used to prevent pods from being scheduled on certain nodes unless they explicitly "tolerate" the node’s taint. Taints are applied to nodes, and tolerations are applied to pods to allow them to be scheduled on tainted nodes.

Exercises

1. Create a Pod with Node Affinity

• Create a pod with nginx as the image.
• Add a Node Affinity rule with the property requiredDuringSchedulingIgnoredDuringExecution, setting the condition disktype = ssd.

apiVersion: v1
kind: Pod
metadata:
  name: nginx
spec:
  containers:
  - name: nginx
    image: nginx

  affinity:
    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
        - matchExpressions:
          - key: disktype
            operator: In
            values:
            - ssd

2. Check Pod Status

• Check the status of the pod to understand why it’s not being scheduled.

Command:

kubectl get pods -o wide

Output:

NAME    READY   STATUS    RESTARTS   AGE     IP       NODE     NOMINATED NODE   READINESS GATES
nginx   0/1     Pending   0          3m55s   <none>   <none>   <none>           <none>

Explanation:

The issue is that we haven't labeled any nodes with disktype=ssd.

3. Add Node Label

• Add the label disktype=ssd to the node kind-cka-cluster-worker.
• Verify the pod status again to ensure it has been scheduled on worker01.

Command to Add the Label:

kubectl label nodes kind-cka-cluster-worker disktype=ssd

Recheck Pod Status:

To verify the change, run:

kubectl get pods -o wide

You should see the nginx pod in the Running state, with the NODE column showing kind-cka-cluster-worker.

4. Create a Second Pod with Node Affinity

• Create a new pod configuration with redis as the image.
• Add a Node Affinity rule with the property requiredDuringSchedulingIgnoredDuringExecution, setting disktype as the condition without specifying a value.

To accomplish this:

   cat pod.yml > redis.yml  

or

Edit the redis.yml file:

• Change the name to redis.
• Update the image to redis.
• Modify the nodeAffinity rule by removing the values section and changing the operator to Exists.

The final configuration should look like this:

apiVersion: v1
kind: Pod
metadata:
  name: redis
spec:
  containers:
  - name: redis
    image: redis
  affinity:
    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
        - matchExpressions:
          - key: disktype
            operator: Exists

Do not deploy this pod yet—we will first ensure that the cluster is prepared for this configuration.

Add the Label disktype (with no value) to worker02 Node:

To add a label with just the key disktype and no value to the node kind-cka-cluster-worker2, run the following command:

   kubectl label nodes kind-cka-cluster-worker2 disktype=

Verify the Node Labels:

Check if the label has been successfully added to the node:

   kubectl get nodes --show-labels

Look for the disktype label under LABELS for kind-cka-cluster-worker2.

Deploy the Redis Pod:

Ensure that the redis.yml file is configured with the nodeAffinity rule that requires the disktype key to exist.

Deploy the pod:

   kubectl apply -f redis.yml

Check Pod Scheduling:

Verify that the redis pod is now scheduled on worker02:

   kubectl get pods -o wide

The output should show kind-cka-cluster-worker2 in the NODE column for the redis pod.

Tags and Mentions

• @piyushsachdeva
• Day 15: Video Tutorial