Project 5 — Master–Agent Architecture

Why Master–Agent?

As your team grows, you distribute builds across machines.

GitHub Push
   ↓
Jenkins Master
   ↓
Agent (EC2)
   ↓
Build + Test + Docker + Deploy

Step 1 — Launch an agent EC2

1. Launch another Ubuntu EC2 (agent)
2. Attach Elastic IP if useful for labs
3. Security group: allow SSH from the Jenkins master (not 0.0.0.0/0 in production)

Step 2 — Prepare the agent

On the agent EC2:

sudo apt update
sudo apt install -y docker.io nodejs npm
 
sudo systemctl start docker
sudo systemctl enable docker
sudo usermod -aG docker ubuntu
newgrp docker

Verify:

node -v
npm -v
docker -v

Step 3 — Add the agent in Jenkins

1. Manage Jenkins → Nodes → New Node

2. Node name: agent-ec2
3. Type: Permanent Agent → Create

Configure:

Save. Jenkins attempts to connect.

Step 4 — Verify

Manage Jenkins → Nodes should show:

agent-ec2 → ONLINE

The label docker-agent is what Jenkinsfiles will target.

Using agents in a Jenkinsfile

Entire pipeline on one agent

pipeline {
  agent { label 'docker-agent' }
 
  stages {
    stage('Build') {
      steps {
        sh 'make build'
      }
    }
  }
}

Different stages on different agents

pipeline {
  agent none
 
  stages {
    stage('Build on Linux') {
      agent { label 'linux' }
      steps { sh 'npm run build' }
    }
 
    stage('Test on Docker') {
      agent {
        docker {
          image 'node:18-alpine'
          label 'docker-agent'
        }
      }
      steps { sh 'npm test' }
    }
 
    stage('Deploy') {
      agent { label 'production' }
      steps { sh './deploy.sh' }
    }
  }
}

Agent launch methods

Hands-on: run a pipeline on the agent

1. Create a Pipeline job (or reuse Project 3)
2. Keep webhook / SCM config as before
3. Set agent { label 'docker-agent' } in the Jenkinsfile (commit + push)
4. Confirm the build runs on agent-ec2 (build page → agent name)