binlab/jenkins-agent repository overview

⁠Jenkins Agent based on Alpine Linux

⁠Current versions included in a latest builds:

Here different versions of Jenkins Agent images for different tasks

⁠JNLP Agent + with Docker client and AWS CLI

• 3.29-jnlp, jnlp, latest (jnlp/Dockerfile)⁠

• 3.29-jnlp-docker, jnlp-docker (jnlp/docker/Dockerfile)⁠

• 3.29-jnlp-docker-awscli, jnlp-docker-awscli (jnlp/docker-awscli/Dockerfile)⁠

  This is an image for using JNLP to establish a connection behind NAT and with built Docker client and + AWS CLI. Additional for user jenkins added group docker with GID 994 for communication with host Docker by socket /var/run/docker.sock without root permissions.

• 3.29-jnlp-docker-awscli-kubectl-helm, jnlp-docker-awscli-kubectl-helm (jnlp/docker-awscli-kubectl-helm/Dockerfile)⁠

  This image is similar to the jnlp-docker-awscli image but has Kubernetes specific tools kubectl and helm

• 3.29-jnlp-docker-awscli-kubectl-helm-kubernetes-deploy-kapp, jnlp-docker-awscli-kubectl-helm-kubernetes-deploy-kapp (jnlp/docker-awscli-kubectl-helm/kubernetes-deploy-kapp/Dockerfile)⁠

  This image is similar to the jnlp-docker-awscli-kubectl-helm image but has additional Kubernetes tools kubernetes-deploy and kapp

  Running by Docker:

  docker run binlab/jenkins-agent:jnlp \
      -url "https://jenkins.example.com" \
      -workDir=/home/jenkins/agent \
      <secret> <agent name>
  

  Copy

  Running by Docker Compose:

  version: "3.3"
  services:
    jenkins-agent:
      image: binlab/jenkins-agent:jnlp-docker
      container_name: jenkins-agent
      hostname: jenkins-agent
      restart: unless-stopped
      environment:
        JENKINS_URL: "https://jenkins.example.com"
        JENKINS_AGENT_WORKDIR: "/home/jenkins/agent"
        JENKINS_SECRET: "jenkins-agent-secret"
        JENKINS_AGENT_NAME: "project-agent"
      volumes:
        - jenkins-agent:/home/jenkins/agent:rw
        - /var/run/docker.sock:/var/run/docker.sock:rw
      networks:
        - jenkins-agent
      logging:
        driver: "json-file"
        options:
          max-size: "10m"
          max-file: "3"
  
  volumes:
    jenkins-agent:
      driver: local
  
  networks:
    jenkins-agent:
      driver: bridge
  

  Copy

  Optional environment variables:

• JENKINS_URL: url for the Jenkins server, can be used as a replacement to -url option, or to set alternate jenkins URL

• JENKINS_TUNNEL: (HOST:PORT) connect to this agent host and port instead of Jenkins server, assuming this one do route TCP traffic to Jenkins master. Useful when when Jenkins runs behind a load balancer, reverse proxy, etc.

• JENKINS_SECRET: agent secret, if not set as an argument

• JENKINS_AGENT_NAME: agent name, if not set as an argument

• JENKINS_AGENT_WORKDIR: agent work directory, if not set by optional parameter -workDir

⁠Configuration specifics

⁠Enabled JNLP protocols

By default, the JNLP3-connect⁠ is disabled due to the known stability and scalability issues. You can enable this protocol on your own risk using the JNLP_PROTOCOL_OPTS=-Dorg.jenkinsci.remoting.engine.JnlpProtocol3.disabled=false property (the protocol should be enabled on the master side as well).

In Jenkins versions starting from 2.27 there is a JNLP4-connect⁠ protocol. If you use Jenkins 2.32.x LTS, it is recommended to enable the protocol on your instance.

⁠Amazon ECS

Make sure your ECS container agent is updated⁠ before running. Older versions do not properly handle the entryPoint parameter. See the entryPoint⁠ definition for more information.

⁠Jnlp auto version (see jnlp-auto folder)

This jenkins-agent version can automatically register in the jenkins server.

⁠Permissions on the jenkins server

The required permissions for the user are:

• Overall/Read
• Agent/Connect
• Agent/Create
• Agent/Delete

⁠Additional ebvironment variables

• AUTO_REG: tell jenkins-slave script that we want to automatically register our agent. just delete it if you don't want to use it.
• ENV: setup environment variable (also, set a label for the slave agent)

⁠Docker example

version: "3.3"
services:
  jenkins-agent:
    image: binlab/jenkins-agent:jnlp-docker
    container_name: jenkins-agent
    hostname: jenkins-agent
    restart: unless-stopped
    environment:
      JENKINS_URL: "https://jenkins.example.com"
      JENKINS_AGENT_WORKDIR: "/home/jenkins/agent"
      JENKINS_SECRET: "example-bot:296682da3b9c6dc624551b1e17c"
      JENKINS_AGENT_NAME: "project-agent"
      AUTO_REG: "true"
      ENV: "dev"

⁠Entrypoint

Because of using sh trap we need to install dumb-init and set it up as entrypoint