Assigning a Kubernetes Pod to a Specific Node

Introduction

Kubernetes scheduling usually doesn’t need much help in order to determine which node to run a pod on. However, you may occasionally wish to have a little more control. In this lab, you will be able to practice the process of ensuring a pod runs on a specific node.

Solution

Log in to the lab server using the credentials provided:

ssh cloud_user@<PUBLIC_IP_ADDRESS>

Configure the auth-gateway Pod to Only Run on k8s-worker2

1. Attach a label to k8s-worker2.

   kubectl label nodes k8s-worker2 external-auth-services=true
   

2. Open auth-gateway.yml:

   vi auth-gateway.yml
   

3. Add a nodeSelector to the auth-gateway pod descriptor:

   ...
   
   spec:
     nodeSelector:
       external-auth-services: "true"
   
     ...
   

4. Save and exit the file by pressing Escape followed by :wq.

5. Delete and re-create the pod:

   kubectl delete pod auth-gateway -n beebox-auth
   

   kubectl create -f auth-gateway.yml
   

6. Verify the pod is scheduled on the k8s-worker2 node:

   kubectl get pod auth-gateway -n beebox-auth -o wide
   

Configure the auth-data Deployment’s Replica Pods to Only Run on k8s-worker2

1. Open auth-data.yml:

   vi auth-data.yml
   

2. Add a nodeSelector to the pod template in the deployment spec (it will be the second spec in the file):

   ...
   
   spec:
   
     ...
   
     template:
   
       ...
   
       spec:
         nodeSelector:
           external-auth-services: "true"
   
         ...
   

3. Save and exit the file by pressing Escape followed by :wq.

4. Update the deployment:

   kubectl apply -f auth-data.yml
   

5. Verify the deployment’s replicas are all running on k8s-worker2:

   kubectl get pods -n beebox-auth -o wide