How I Deployed Jenkins on Kubernetes Using Dynamic Agent Pods

In this article, I’ll walk you through how I deployed Jenkins on Kubernetes, configured dynamic agent pods, and successfully executed a pipeline using Kubernetes-native Jenkins agents.

This setup reflects real-world CI/CD architecture where Jenkins scales automatically using Kubernetes pods instead of static agents.

Why Jenkins on Kubernetes?

Traditional Jenkins agents are:

• Hard to scale

• Costly when idle

• Difficult to manage

With Kubernetes:

• Agent pods are created on demand

• Pods terminate automatically after job completion

• Better resource utilization

• Cloud-native CI/CD

Architecture Overview

• Jenkins Master runs as a Kubernetes pod

• Kubernetes Plugin provisions agent pods dynamically

• NodePort Service exposes Jenkins UI

• ServiceAccount + RBAC allows Jenkins to manage pods

Step 1: Deploy Jenkins Master on Kubernetes

I deployed Jenkins using a Deployment manifest with:

• jenkins/jenkins:lts-jdk17 image

• Exposed ports 8080 (UI) and 50000 (JNLP agents)

• emptyDir volume for Jenkins home

This ensured Jenkins runs reliably inside the cluster.

Step 2: Expose Jenkins Using NodePort

To access Jenkins from the browser, I created a NodePort service:

• Jenkins UI → NodeIP:30000

• Agent communication → 50000

This made Jenkins accessible externally while keeping things simple for demo purposes.

Step 3: ServiceAccount and RBAC Configuration

Since Jenkins needs to:

• Create pods

• Delete pods

• Watch Kubernetes resources

I created:

• A ServiceAccount

• A ClusterRoleBinding with cluster-admin permissions (for learning/demo)

  In production, permissions should always follow the least privilege principle.

Step 4: Configure Kubernetes Cloud in Jenkins

Inside Jenkins UI:

Manage Jenkins → Configure System → Clouds → Kubernetes

Configured:

• Kubernetes URL: https://kubernetes.default:443

• Jenkins URL: http://jenkins:8080

• Jenkins Tunnel: jenkins:50000

• Credentials: Kubernetes Service Account

• Namespace: default

This allows Jenkins to talk directly to the Kubernetes API.

Step 5: Create Jenkins Agent Pod Template

Configured an agent pod template with:

• Label: jenkins-agent

• Image: jenkins/inbound-agent:lts-jdk17

• Container name: jnlp

• Working directory: /home/jenkins/agent

This template is used whenever a pipeline requests an agent with this label.

Step 6: Run a Pipeline Using Kubernetes Agent

Sample pipeline:

pipeline {
agent { label ‘jenkins-agent’ }
stages {
stage(‘Test’) {
steps {
sh ‘echo Hello from Kubernetes Jenkins Agent’
}
}
}
}

What happens internally?

• Jenkins requests an agent

• Kubernetes creates a new pod

• Pipeline runs inside the pod

• Pod terminates after completion

Final Outcome

Jenkins master running on Kubernetes
Agent pods created dynamically
Pipeline executed successfully
Pods cleaned up automatically
Jenkins UI accessible via NodePort

Key Takeaways

• Kubernetes-based Jenkins is scalable and efficient

• No idle agents wasting resources

• Best approach for cloud-native CI/CD

• Strongly recommended for DevOps engineers

Who Should Read This?

• DevOps Engineers

• Kubernetes Beginners

• CI/CD Learners

• Jenkins Interview Preparation

Conclusion

Running Jenkins on Kubernetes with dynamic agent pods is a modern DevOps best practice. This setup not only improves scalability but also simplifies agent management dramatically.

If you’re preparing for DevOps interviews or real-world CI/CD projects, this architecture is a must-know.