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’ll learn how to run Jenkins build agents inside Kubernetes Pods using a Declarative Pipeline. We assume you have already installed and configured the Kubernetes plugin for Jenkins.

​
Prerequisites

  • A running Jenkins instance with the Kubernetes plugin installed
  • Access to a Kubernetes cluster and proper credentials configured under Manage Jenkins → Configure System → Cloud → Kubernetes
Make sure your Jenkins service account has permissions to create and delete Pods in the target namespace.

​
1. Create a New Pipeline Job

  1. Open the Jenkins Dashboard and click New Item.
  2. Enter a name, for example k8s-cloud-agent-demo, and select Pipeline.
  3. Scroll to the Pipeline section, choose Pipeline script, then pick the Declarative Kubernetes sample.
The image shows a Jenkins interface where a user is creating a new item named "k8s-cloud-agent-demo." Various project types like Freestyle project, Pipeline, Multi-configuration project, and Folder are listed as options.

​
2. Define a Basic Kubernetes Agent

The default Declarative example includes an agent block with containerTemplate. To get started quickly, simplify it to print the Pod’s hostname: 
pipeline {
  agent { kubernetes {
    containerTemplate {
      name 'shell'
      image 'ubuntu'
      command 'sleep'
      args 'infinity'
    }
  } }
  stages {
    stage('Print Hostname') {
      steps {
        sh 'hostname'
        sh 'sleep 120s'
      }
    }
  }
}
To focus further, remove the containerTemplate and let Jenkins use defaults: 
pipeline {
  agent { kubernetes { } }
  stages {
    stage('Print Hostname') {
      steps {
        sh 'hostname'
        sh 'sleep 120s'
      }
    }
  }
}
By default, Pods are deleted after the build completes (podRetention: Never).
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.
With the default retention policy (Never), Pods are removed immediately after each build. Disable or adjust this if you need to debug post-build.

​
3. Inline Pod Template with YAML

For more control, define your Pod in YAML directly within the pipeline: 
pipeline {
  agent {
    kubernetes {
      yaml '''
apiVersion: v1
kind: Pod
spec:
  containers:
  - name: ubuntu-container
    image: ubuntu
    command:
    - sleep
    args:
    - infinity
'''
    }
  }
  stages {
    stage('Print Hostname') {
      steps {
        sh 'hostname'
      }
    }
  }
}
You can also reference an external Pod template file in multibranch pipelines: 
agent {
  kubernetes {
    yamlFile 'jenkins-pod.yaml'
  }
}
Global Pod templates can be managed under Manage Jenkins → Kubernetes.
The image shows a Jenkins interface for managing pod templates, with an option to add a new pod template.

​
4. Run the Pipeline

  1. Save the Pipeline script.
  2. Click Build Now.
Jenkins will provision a Kubernetes Pod based on your definition. In the console output, look for Pod creation logs and your hostname: 
[Pipeline] sh
jenkins-agent-abc123
[Pipeline] // sh
The image shows a Jenkins console output screen displaying details of a pipeline execution, including the creation of a Kubernetes pod.

​
5. Default Cloud & Advanced Configuration

If you don’t specify a cloud in the agent block, Jenkins selects the first available Kubernetes cloud. You can view options in the Declarative Directive Generator: 
agent {
  kubernetes {
    // ...
  }
}
The image shows a Jenkins interface with the "Declarative Directive Generator" open, allowing users to generate pipeline code. A dropdown menu is visible, offering options for running agents, including Docker and Kubernetes.

​
Select a Specific Cloud and Namespace

pipeline {
  agent {
    kubernetes {
      cloud 'dasher-prod-k8s-us-east'
      yaml '''
apiVersion: v1
kind: Pod
spec:
  containers:
  - name: ubuntu-container
    image: ubuntu
    command:
    - sleep
    args:
    - infinity
'''
    }
  }
  stages {
    stage('Print Hostname') {
      steps {
        sh 'hostname'
      }
    }
  }
}
The image shows a Jenkins configuration interface for setting up a Kubernetes agent, with options for selecting the cloud, namespace, and default container.

​
6. Example: Multiple Containers

Define multiple containers in the same Pod to run different tools per stage: 
pipeline {
  agent {
    kubernetes {
      cloud 'dasher-prod-k8s-us-east'
      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'
        }
      }
    }
  }
}
If you omit container('node-container'), the Node.js commands default to ubuntu-container and will fail.
The image shows a Jenkins dashboard with a pipeline named "k8s-cloud-agent-demo," displaying the status of two builds with stages like "Print Hostname" and "Print Node Version." The sidebar includes options like "Build Now," "Configure," and "Open Blue Ocean."

​
7. Inspecting Pods and Events

During a build, you can inspect Pods and events with kubectl: 
# List Jenkins agent Pods
kubectl -n jenkins get pods

# View Pod lifecycle events
kubectl -n jenkins get events
You’ll see scheduling, image pull, start, and termination events for each container.

​
Conclusion

Leveraging Kubernetes Pods as Jenkins agents in Declarative Pipelines offers:
  • Custom environments per stage (e.g., Ubuntu, Node.js)
  • Ephemeral build agents that spin up and down automatically
  • Inline YAML definitions or external templates for reusable configurations
Start defining your CI/CD workloads with fine-grained control over tools and dependencies!

Watch Video