Roll Update and Roll back with Kubernetes

Project: student-feedback-app

This is a simple Flask application that:

• Shows Application Version

• Shows Environment Mode

• Runs inside Docker

• Deployed in Kubernetes

• Uses Rolling Update

• Uses ConfigMap

Everything with real names.

PART 1 – Create Real Application

Step 1: Create Project Folder

On your system:

mkdir student-feedback-app
cd student-feedback-app

Step 2: Create Flask Application

Create file:

app.py

from flask import Flask
import os

app = Flask(__name__)

@app.route("/")
def home():
    version = os.getenv("APP_VERSION")
    mode = os.getenv("APP_MODE")
    return f"""
    <h2>Student Feedback Application</h2>
    <p>Application Version: {version}</p>
    <p>Environment Mode: {mode}</p>
    """

if __name__ == "__main__":
    app.run(host="0.0.0.0", port=5000)

What we are doing here?

• os.getenv() reads environment variables

• We are NOT hardcoding version

• We are NOT hardcoding mode

• These values will come from Kubernetes

This is real DevOps practice.

Step 3: Create requirements.txt

flask

Step 4: Create Dockerfile

FROM python:3.9-slim

WORKDIR /app

COPY . .

RUN pip install -r requirements.txt

EXPOSE 5000

CMD ["python", "app.py"]

What each line does?

Line	Meaning
FROM	Base image
WORKDIR	Creates working directory
COPY	Copies project files
RUN	Installs flask
EXPOSE	Opens port 5000
CMD	Starts app

PART 2 – Build Docker Image

Step 5: Build Image

docker build -t student-feedback-app:v1 .

What happens here?

• Docker reads Dockerfile

• Creates image named student-feedback-app

• Version tagged as v1

PART 3 – Deploy to Kubernetes

Step 6: Create Deployment YAML

Create file:

student-feedback-deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: student-feedback-deployment
spec:
  replicas: 3
  selector:
    matchLabels:
      app: student-feedback
  template:
    metadata:
      labels:
        app: student-feedback
    spec:
      containers:
      - name: student-feedback-container
        image: student-feedback-app:v1
        imagePullPolicy: Never
        ports:
        - containerPort: 5000
        env:
        - name: APP_VERSION
          value: "1.0"
        - name: APP_MODE
          value: "development"

Explanation of Important Sections

metadata.name

Real name of deployment:

student-feedback-deployment

replicas: 3

Creates 3 Pods → High Availability

matchLabels

Connects Deployment to Pods

imagePullPolicy: Never

Since we built image locally

env:

We are setting:

APP_VERSION = 1.0
APP_MODE = development

These go inside container as environment variables.

Step 7: Apply Deployment

kubectl apply -f student-feedback-deployment.yaml

Check:

kubectl get pods

You should see 3 running pods.

PART 4 – Expose Application

Step 8: Create Service

kubectl expose deployment student-feedback-deployment --type=NodePort --port=5000

Check service:

kubectl get svc

Open:

http://localhost:<nodeport>

You will see:

Student Feedback ApplicationApplication Version: 1.0Environment Mode: development

PART 5 – Rolling Update (Real Example)

Now imagine we improved UI and want version 2.

Step 9: Modify app.py

Change heading to:

<h2>Student Feedback Application - Updated UI</h2>

Build new image:

docker build -t student-feedback-app:v2 .

Step 10: Update Deployment Image

kubectl set image deployment/student-feedback-deployment \
student-feedback-container=student-feedback-app:v2

What happens?

• Kubernetes does NOT delete all pods

• It replaces pods one by one

• Zero downtime

• This is Rolling Update

Check:

kubectl rollout status deployment student-feedback-deployment

Rollback (If Something Breaks)

kubectl rollout undo deployment student-feedback-deployment

Kubernetes returns to previous version automatically.

This is real production practice.

PART 6 – Use ConfigMap (Real DevOps Practice)

Now instead of hardcoding APP_MODE inside YAML, we externalize it.

Step 11: Create ConfigMap

kubectl create configmap student-feedback-config \
--from-literal=APP_MODE=production

Check:

kubectl get configmap

Step 12: Update Deployment YAML

Replace APP_MODE section:

- name: APP_MODE
  valueFrom:
    configMapKeyRef:
      name: student-feedback-config
      key: APP_MODE

Apply again:

kubectl apply -f student-feedback-deployment.yaml

What happened?

• We did NOT rebuild Docker image

• We changed only configuration

• App now shows:

Environment Mode: production

That is clean DevOps architecture.