Deployments and ReplicaSets

# CHAPTER 6

Deployments and ReplicaSets #

1. Introduction #

2. Learning Objectives #

• Differentiate between a Pod, a ReplicaSet, and a Deployment.

• Understand how a ReplicaSet ensures Self-Healing and Scalability.

• Author a Deployment YAML manifest.

• Execute a Zero-Downtime Rolling Update to a new application version.

• Execute a Rollback if a deployment fails.

3. Beginner-Friendly Explanation #

• The Pod: A single cashier. If the cashier gets sick and goes home, the register is empty.

• The ReplicaSet: The Assistant Manager. Their *only* job is to ensure there are always exactly 3 cashiers working. If a cashier gets sick, the Assistant Manager instantly hires a new one to maintain the count of 3.

• The Deployment: The General Manager. When the restaurant changes uniforms from Red to Blue, the General Manager doesn't fire all 3 cashiers at once (causing downtime). They tell the Assistant Manager: "Hire 1 new cashier in a Blue uniform. Once they are working, fire 1 Red cashier. Repeat until everyone is Blue."

*A Deployment manages a ReplicaSet. A ReplicaSet manages Pods.*

4. What is a ReplicaSet? #

• If it sees 2 Pods, it creates 1.

• If it sees 4 Pods (maybe you manually created one by mistake), it ruthlessly assassinates 1 to maintain the exact desired state of 3.

5. What is a Deployment? #

When you update your application's Docker image from v1 to v2, the Deployment orchestrates a Rolling Update:

1. 1. It creates a brand new, secondary ReplicaSet for v2.

1. 2. It spins up a v2 Pod.

1. 3. It gracefully terminates a v1 Pod.

1. 4. It repeats this process one by one until the v1 ReplicaSet is scaled to 0, and the v2 ReplicaSet is scaled to your desired capacity.

6. Anatomy of a Deployment YAML #

apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-webapp-deployment
spec:
  replicas: 3
  selector:
    matchLabels:
      app: webapp
  template: # This is the blueprint for the Pods the Deployment will create
    metadata:
      labels:
        app: webapp
    spec:
      containers:
      - name: nginx
        image: nginx:1.24
        ports:
        - containerPort: 80

7. Mini Project: Scale and Update an App #

Step-by-Step Tutorial:

1. 1. Save the YAML from Section 6 into a file named deployment.yaml.

1. 2. Apply the blueprint:

kubectl apply -f deployment.yaml

1. 3. Check the status. You will see 1 Deployment, 1 ReplicaSet, and 3 Pods!

kubectl get deployments
kubectl get rs
kubectl get pods

1. 4. Test Self-Healing: Manually delete one of your Pods (kubectl delete pod <pod-name>). Immediately run kubectl get pods again. You will see Kubernetes instantly creating a new Pod to replace the murdered one!

1. 5. Test Scaling: We need to handle a traffic spike! Scale the deployment to 5 replicas:

kubectl scale deployment my-webapp-deployment --replicas=5

1. 6. Test Rolling Update: Let's upgrade our Nginx version from 1.24 to 1.25:

kubectl set image deployment/my-webapp-deployment nginx=nginx:1.25

1. 7. Watch the rollout happen live!

kubectl rollout status deployment/my-webapp-deployment

8. Real-World Scenarios #

9. Best Practices #

• Never Modify Pods Managed by a Deployment: If you kubectl exec into a Pod managed by a deployment and manually update a configuration file, your changes will be destroyed the next time the Pod restarts. Always update the deployment.yaml file and let the controller handle the rollout.

10. Common Mistakes #

• Selector Mismatches: A Deployment uses spec.selector.matchLabels to find the Pods it is supposed to manage. If this selector does not EXACTLY match the labels defined in spec.template.metadata.labels, the Deployment will fail to create, throwing cryptic errors about label mismatches.

11. Exercises #

1. 1. If you run kubectl get rs, you often see multiple ReplicaSets for the same application, but all of them except one have a DESIRED count of 0. Explain why.

1. 2. Outline the steps to perform an emergency rollback of a Deployment to a previous version.

12. FAQs #

13. Interview Questions #

• Q: Explain the hierarchy and relationship between a Deployment, a ReplicaSet, and a Pod. Why does Kubernetes utilize this layered controller architecture?

• Q: Detail the mechanics of a Kubernetes Rolling Update. How does the Deployment controller guarantee that an application remains available to users during a version transition?

14. Summary #

15. Next Chapter Recommendation #