How to Create Static and Dynamic Jenkins Agents on Kubernetes

Manually create a Jenkins agent node on the Jenkins server by selecting the JNLP type and retrieve the required JENKINS_AGENT_NAME and JENKINS_AGENT_SECRET values.

For a static agent, write a Deployment manifest that includes the agent name, namespace, container image, resource limits, and environment variables for the Jenkins URL, secret, and name, then apply it with kubectl apply.

For a dynamic agent, install the Kubernetes plugin, configure a cloud with the cluster information, and define the pipeline agent section using a YAML pod template and the necessary stages.

Steps to create the node in the Jenkins UI (Node Management → New Node) are illustrated with screenshots, showing the configuration of node name and secret, which are later used in the deployment.

Static agent deployment example:

kind: Deployment
apiVersion: apps/v1
metadata:
  labels:
    k8s-app: jenkinsagent
  name: jenkinsagent
  namespace: devops
spec:
  replicas: 1
  revisionHistoryLimit: 10
  selector:
    matchLabels:
      k8s-app: jenkinsagent
  template:
    metadata:
      labels:
        k8s-app: jenkinsagent
    namespace: devops
    name: jenkinsagent
    spec:
      containers:
        - name: jenkinsagent
          image: jenkins/inbound-agent:4.10-3-jdk8
          securityContext:
            privileged: true
          imagePullPolicy: IfNotPresent
          resources:
            limits:
              cpu: 1000m
              memory: 2Gi
            requests:
              cpu: 500m
              memory: 512Mi
          env:
            - name: JENKINS_URL
              value: http://192.168.1.200:8080
            - name: JENKINS_SECRET
              value: 2f3f27ee602171c5b51bc69c2bda501d703ac65f42de16600b3e3007dfdec422
            - name: JENKINS_AGENT_NAME
              value: buildserver1
            - name: JENKINS_AGENT_WORKDIR
              value: /home/jenkins/workspace

Dynamic agent pipeline (using the Kubernetes plugin) example:

@Library("mylib@feature-k8s") _
podYaml = """
kind: Pod
apiVersion: v1
metadata:
  labels:
    k8s-app: jenkinsagent
  name: jnlp
  namespace: devops
spec:
  securityContext:
    runAsUser: 0
  containers:
    - name: jnlp
      image: jenkins/inbound-agent:4.10-3-jdk8
      imagePullPolicy: IfNotPresent
    - name: maven
      image: maven:3.8.4-jdk-8
      imagePullPolicy: IfNotPresent
      command:
      - cat
      tty: true
    - name: dind
      image: 'docker:stable-dind'
      command:
      - dockerd
      - --host=unix:///var/run/docker.sock
      - --host=tcp://0.0.0.0:8000
      - --insecure-registry=192.168.1.200:8088
      securityContext:
        privileged: true
      volumeMounts:
      - mountPath: /var/run
        name: docker-dir
    - name: "docker"
      tty: true
      image: docker:19.03.15-git
      imagePullPolicy: IfNotPresent
      volumeMounts:
      - mountPath: /var/run
        name: docker-dir
    - name: "curl"
      tty: true
      image: curlimages/curl:7.84.0
      imagePullPolicy: IfNotPresent
      command:
      - cat
    - name: "helm"
      tty: true
      image: dtzar/helm-kubectl:3.9.0
      imagePullPolicy: IfNotPresent
      command:
      - cat
  volumes:
    - name: docker-dir
      emptyDir: {}
"""

pipeline {
  agent {
    kubernetes {
      label "test01"
      cloud 'kubernetes'
      yaml podYaml
    }
  }
  options { skipDefaultCheckout() }
  stages {
    stage("Checkout") {
      steps {
        script {
          println "GetCode"
          // checkout code and set env variables
        }
      }
    }
    // Additional stages: Build, PushArtifact, Docker, ReleaseFile, HelmDeploy, CheckHealth, RollOut
  }
}

// Helper functions: GetCommitID, PushArtifact, HttpReqByHttpRequest, HttpReq, etc.

The pipeline includes stages for checking out source code, building with Maven, pushing artifacts to a Nexus repository, building and pushing Docker images, updating Helm values files in GitLab, deploying with Helm, health checking, and optional rollback via user input.