Skip to main content

Documentation Index

Fetch the complete documentation index at: https://notes.kodekloud.com/llms.txt

Use this file to discover all available pages before exploring further.

In this guide, you will learn how to configure a Jenkins Declarative Pipeline to dynamically spin up Kubernetes Pods as build agents. By the end, you can:
  • Define Pod templates inline or via YAML files
  • Use a default container or target specific containers per stage
  • Inspect Pod definitions and events on your cluster

​
1. Create a New Pipeline Job

  1. From the Jenkins dashboard, click New Item.
The image shows a Jenkins dashboard interface with a menu open, displaying options like "New Item" and "Build History." There is also a section for managing clouds with a button to add a new cloud.
  1. Enter the job name (k8s-cloud-agent-demo or your preferred name) and select Pipeline.
The image shows a Jenkins interface where a user is creating a new item named "k8s-cloud-agent-demo." Various project types like Freestyle, Pipeline, and Multi-configuration are listed as options.
  1. Scroll to the Pipeline section, choose Pipeline script, and prepare to paste your Declarative Pipeline.

​
2. Basic Declarative Pipeline with Kubernetes Agent

Start with a minimal pipeline that runs one stage inside a Kubernetes Pod defined inline: 
pipeline {
  agent {
    kubernetes {
      yaml '''
apiVersion: v1
kind: Pod
spec:
  containers:
    - name: ubuntu-container
      image: ubuntu
      command: ["sleep"]
      args: ["infinity"]
  securityContext:
    runAsUser: 1000
'''
    }
  }
  stages {
    stage('Print Hostname') {
      steps {
        sh 'hostname'
        sh 'sleep 120s'  // Keep Pod alive for inspection
      }
    }
  }
}
In the Kubernetes Cloud settings, Pod Retention is set to Never, so Pods are deleted after each build.
The image shows a configuration screen for a Jenkins cloud setup, with options for pod retention, maximum connections to the Kubernetes API, and other settings.

​
3. Run the Build and Inspect the Pod

  1. Click Build Now on your Pipeline job.
  2. Open Console Output to see the steps: 
    [Pipeline] sh
+ hostname
k8s-cloud-agent-demo-1-36qw9-qdg7z-3mc58
[Pipeline] sh
+ sleep 120s
  3. Jenkins also prints the full Pod spec it created: 
    ---
apiVersion: "v1"
kind: "Pod"
metadata:
  name: "k8s-cloud-agent-demo-1-36qw9-qdg7z-3mc58"
  namespace: "jenkins"
  labels:
    jenkins/label: "k8s-cloud-agent-1-36qw9"
spec:
  containers:
    - name: ubuntu-container
      image: ubuntu
      command: ["sleep"]
      args: ["infinity"]
  4. On your Kubernetes cluster, verify the Pod: 
    kubectl -n jenkins get pod
kubectl -n jenkins describe pod k8s-cloud-agent-demo-1-36qw9-qdg7z-3mc58
The image shows a terminal window displaying Kubernetes pod information, including volume details and event logs related to container creation and scheduling.
  1. You will notice two containers:
    • ubuntu-container (your build environment)
    • jnlp (inbound Jenkins agent for communication)
    containers:
  - name: ubuntu-container
    image: ubuntu
    command: [sleep]
    args: [infinity]
  - name: jnlp
    image: jenkins/inbound-agent:latest
    # ...additional fields...
  2. After 120 seconds, the Pod is removed due to Pod Retention: Never.

​
4. Viewing Pod Events

To review lifecycle events of recently terminated Pods: 
kubectl -n jenkins get events --sort-by='.lastTimestamp'
You’ll see events such as Scheduled, Pulling, Started, and Killing.

​
5. Using Multiple Containers

Demonstrate how to define and select containers per stage.

​
5.1 Define Two Containers with a Default

pipeline {
  agent {
    kubernetes {
      yaml '''
apiVersion: v1
kind: Pod
spec:
  containers:
    - name: ubuntu-container
      image: ubuntu
      command: ["sleep"]
      args: ["infinity"]
    - name: node-container
      image: node:18-alpine
      command: ["cat"]
      tty: true
'''
      defaultContainer 'ubuntu-container'
    }
  }
  stages {
    stage('Print Hostname') {
      steps {
        sh 'hostname'
      }
    }
    stage('Print Node Version') {
      steps {
        // This will run in ubuntu-container and fail
        sh 'node -v'
        sh 'npm -v'
      }
    }
  }
}
The Print Node Version stage will fail because it runs in the Ubuntu container, which lacks Node.js.

​
5.2 Select the Node Container for a Specific Stage

Wrap Node.js commands in a container('node-container') block: 
pipeline {
  agent {
    kubernetes {
      yaml '''
apiVersion: v1
kind: Pod
spec:
  containers:
    - name: ubuntu-container
      image: ubuntu
      command: ["sleep"]
      args: ["infinity"]
    - name: node-container
      image: node:18-alpine
      command: ["cat"]
      tty: true
'''
    }
  }
  stages {
    stage('Print Hostname') {
      steps {
        sh 'hostname'
      }
    }
    stage('Print Node Version') {
      steps {
        container('node-container') {
          sh 'node -v'
          sh 'npm -v'
        }
      }
    }
  }
}
Run again and check the console: 
[Pipeline] sh
+ hostname
k8s-cloud-agent-demo-4-lhzbb-ss3rk-sp3q6
...
[Pipeline] container
[Pipeline] sh
+ node -v
v18.20.4
[Pipeline] sh
+ npm -v
10.7.0
The image shows a Jenkins console output screen displaying details of a pipeline execution, including information about a created Kubernetes pod.

​
6. Advanced Options

FeatureDescriptionExample
Select CloudChoose a specific Kubernetes cloud configurationcloud 'dasher-prod-k8s-us-east'
External YAML DefinitionUse a YAML file instead of inline yaml blockyamlFile 'jenkins-pod.yaml'
Resource Requests & LimitsDefine CPU/memory requests and limits per containerAdd resources { requests { cpu '100m' } }
Node SelectorsTarget specific nodes in your clusternodeSelector 'disktype': 'ssd'
Retry LogicConfigure retry behavior on failureretries 3
Explore more options with the Declarative Directive Generator.
The image shows a Jenkins interface with the "Declarative Directive Generator" open, allowing users to generate pipeline code for a Declarative Pipeline directive. A dropdown menu is visible with options for running agents, including Docker and Kubernetes.

Watch Video