Chapter 12. Using images

You can manage Jenkins authentication in two ways:

The Jenkins server can be configured with the following environment variables:

If you are going to run Jenkins somewhere other than your same project, you must provide an access token to Jenkins to access your project.

The Jenkins image can be run with mounted volumes to enable persistent storage for the configuration:

To customize the official OpenShift Container Platform Jenkins image, you can use the image as a source-to-image (S2I) builder.

You can use S2I to copy your custom Jenkins jobs definitions, add additional plugins, or replace the provided config.xml file with your own, custom, configuration.

To include your modifications in the Jenkins image, you must have a Git repository with the following directory structure:

pluginId:pluginVersion

The contents of the configuration/ directory is copied to the /var/lib/jenkins/ directory, so you can also include additional files, such as credentials.xml, there.

apiVersion: build.openshift.io/v1
kind: BuildConfig
metadata:
  name: custom-jenkins-build
spec:
  source:                       1
    git:
      uri: https://github.com/custom/repository
    type: Git
  strategy:                     2
    sourceStrategy:
      from:
        kind: ImageStreamTag
        name: jenkins:2
        namespace: openshift
    type: Source
  output:                       3
    to:
      kind: ImageStreamTag
      name: custom-jenkins:latest

The OpenShift Container Platform Jenkins image includes the pre-installed Kubernetes plugin that allows Jenkins agents to be dynamically provisioned on multiple container hosts using Kubernetes and OpenShift Container Platform.

To use the Kubernetes plugin, OpenShift Container Platform provides images that are suitable for use as Jenkins agents including the Base, Maven, and Node.js images.

Both the Maven and Node.js agent images are automatically configured as Kubernetes pod template images within the OpenShift Container Platform Jenkins image configuration for the Kubernetes plugin. That configuration includes labels for each of the images that can be applied to any of your Jenkins jobs under their Restrict where this project can be run setting. If the label is applied, jobs run under an OpenShift Container Platform pod running the respective agent image.

The Jenkins image also provides auto-discovery and auto-configuration of additional agent images for the Kubernetes plugin.

With the OpenShift Container Platform sync plugin, the Jenkins image on Jenkins startup searches for the following within the project that it is running or the projects specifically listed in the plugin’s configuration:

When it finds an image stream with the appropriate label, or image stream tag with the appropriate annotation, it generates the corresponding Kubernetes plugin configuration so you can assign your Jenkins jobs to run in a pod that runs the container image that is provided by the image stream.

The name and image references of the image stream or image stream tag are mapped to the name and image fields in the Kubernetes plugin pod template. You can control the label field of the Kubernetes plugin pod template by setting an annotation on the image stream or image stream tag object with the key agent-label. Otherwise, the name is used as the label.

When it finds a config map with the appropriate label, it assumes that any values in the key-value data payload of the config map contains Extensible Markup Language (XML) that is consistent with the configuration format for Jenkins and the Kubernetes plugin pod templates. A key differentiator to note when using config maps, instead of image streams or image stream tags, is that you can control all the parameters of the Kubernetes plugin pod template.

kind: ConfigMap
apiVersion: v1
metadata:
  name: jenkins-agent
  labels:
    role: jenkins-agent
data:
  template1: |-
    <org.csanchez.jenkins.plugins.kubernetes.PodTemplate>
      <inheritFrom></inheritFrom>
      <name>template1</name>
      <instanceCap>2147483647</instanceCap>
      <idleMinutes>0</idleMinutes>
      <label>template1</label>
      <serviceAccount>jenkins</serviceAccount>
      <nodeSelector></nodeSelector>
      <volumes/>
      <containers>
        <org.csanchez.jenkins.plugins.kubernetes.ContainerTemplate>
          <name>jnlp</name>
          <image>openshift/jenkins-agent-maven-35-centos7:v3.10</image>
          <privileged>false</privileged>
          <alwaysPullImage>true</alwaysPullImage>
          <workingDir>/tmp</workingDir>
          <command></command>
          <args>${computer.jnlpmac} ${computer.name}</args>
          <ttyEnabled>false</ttyEnabled>
          <resourceRequestCpu></resourceRequestCpu>
          <resourceRequestMemory></resourceRequestMemory>
          <resourceLimitCpu></resourceLimitCpu>
          <resourceLimitMemory></resourceLimitMemory>
          <envVars/>
        </org.csanchez.jenkins.plugins.kubernetes.ContainerTemplate>
      </containers>
      <envVars/>
      <annotations/>
      <imagePullSecrets/>
      <nodeProperties/>
    </org.csanchez.jenkins.plugins.kubernetes.PodTemplate>

After it is installed, the OpenShift Container Platform Sync plugin monitors the API server of OpenShift Container Platform for updates to image streams, image stream tags, and config maps and adjusts the configuration of the Kubernetes plugin.

The following rules apply:

To use a container image as a Jenkins agent, the image must run the agent as an entrypoint. For more details about this, refer to the official Jenkins documentation.

If in the config map the <serviceAccount> element of the pod template XML is the OpenShift Container Platform service account used for the resulting pod, the service account credentials are mounted into the pod. The permissions are associated with the service account and control which operations against the OpenShift Container Platform master are allowed from the pod.

Consider the following scenario with service accounts used for the pod, which is launched by the Kubernetes Plugin that runs in the OpenShift Container Platform Jenkins image.

If you use the example template for Jenkins that is provided by OpenShift Container Platform, the jenkins service account is defined with the edit role for the project Jenkins runs in, and the master Jenkins pod has that service account mounted.

The two default Maven and NodeJS pod templates that are injected into the Jenkins configuration are also set to use the same service account as the Jenkins master.

Templates provide parameter fields to define all the environment variables with predefined default values. OpenShift Container Platform provides templates to make creating a new Jenkins service easy. The Jenkins templates should be registered in the default openshift project by your cluster administrator during the initial cluster setup.

The two available templates both define deployment configuration and a service. The templates differ in their storage strategy, which affects whether or not the Jenkins content persists across a pod restart.

To use a PV store, the cluster administrator must define a PV pool in the OpenShift Container Platform deployment.

After you select which template you want, you must instantiate the template to be able to use Jenkins.

In the following example, the openshift-jee-sample BuildConfig object causes a Jenkins Maven agent pod to be dynamically provisioned. The pod clones some Java source code, builds a WAR file, and causes a second BuildConfig, openshift-jee-sample-docker to run. The second BuildConfig layers the new WAR file into a container image.

kind: List
apiVersion: v1
items:
- kind: ImageStream
  apiVersion: image.openshift.io/v1
  metadata:
    name: openshift-jee-sample
- kind: BuildConfig
  apiVersion: build.openshift.io/v1
  metadata:
    name: openshift-jee-sample-docker
  spec:
    strategy:
      type: Docker
    source:
      type: Docker
      dockerfile: |-
        FROM openshift/wildfly-101-centos7:latest
        COPY ROOT.war /wildfly/standalone/deployments/ROOT.war
        CMD $STI_SCRIPTS_PATH/run
      binary:
        asFile: ROOT.war
    output:
      to:
        kind: ImageStreamTag
        name: openshift-jee-sample:latest
- kind: BuildConfig
  apiVersion: build.openshift.io/v1
  metadata:
    name: openshift-jee-sample
  spec:
    strategy:
      type: JenkinsPipeline
      jenkinsPipelineStrategy:
        jenkinsfile: |-
          node("maven") {
            sh "git clone https://github.com/openshift/openshift-jee-sample.git ."
            sh "mvn -B -Popenshift package"
            sh "oc start-build -F openshift-jee-sample-docker --from-file=target/ROOT.war"
          }
    triggers:
    - type: ConfigChange

It is also possible to override the specification of the dynamically created Jenkins agent pod. The following is a modification to the previous example, which overrides the container memory and specifies an environment variable.

kind: BuildConfig
apiVersion: build.openshift.io/v1
metadata:
  name: openshift-jee-sample
spec:
  strategy:
    type: JenkinsPipeline
    jenkinsPipelineStrategy:
      jenkinsfile: |-
        podTemplate(label: "mypod", 1
                    cloud: "openshift", 2
                    inheritFrom: "maven", 3
                    containers: [
            containerTemplate(name: "jnlp", 4
                              image: "openshift/jenkins-agent-maven-35-centos7:v3.10", 5
                              resourceRequestMemory: "512Mi", 6
                              resourceLimitMemory: "512Mi", 7
                              envVars: [
              envVar(key: "CONTAINER_HEAP_PERCENT", value: "0.25") 8
            ])
          ]) {
          node("mypod") { 9
            sh "git clone https://github.com/openshift/openshift-jee-sample.git ."
            sh "mvn -B -Popenshift package"
            sh "oc start-build -F openshift-jee-sample-docker --from-file=target/ROOT.war"
          }
        }
  triggers:
  - type: ConfigChange

By default, the pod is deleted when the build completes. This behavior can be modified with the plugin or within a pipeline Jenkinsfile.

When deployed by the provided Jenkins Ephemeral or Jenkins Persistent templates, the default memory limit is 1 Gi.

By default, all other process that run in the Jenkins container cannot use more than a total of 512 MiB of memory. If they require more memory, the container halts. It is therefore highly recommended that pipelines run external commands in an agent container wherever possible.

And if Project quotas allow for it, see recommendations from the Jenkins documentation on what a Jenkins master should have from a memory perspective. Those recommendations proscribe to allocate even more memory for the Jenkins master.

It is recommended to specify memory request and limit values on agent containers created by the Jenkins Kubernetes plugin. Admin users can set default values on a per-agent image basis through the Jenkins configuration. The memory request and limit parameters can also be overridden on a per-container basis.

You can increase the amount of memory available to Jenkins by overriding the MEMORY_LIMIT parameter when instantiating the Jenkins Ephemeral or Jenkins Persistent template.

The OpenShift Container Platform Jenkins agent images are available on Quay.io or registry.redhat.io.

Jenkins images are available through the Red Hat Registry:

$ docker pull registry.redhat.io/openshift4/ose-jenkins:<v4.5.0>

$ docker pull registry.redhat.io/openshift4/jenkins-agent-nodejs-10-rhel7:<v4.5.0>

$ docker pull registry.redhat.io/openshift4/jenkins-agent-nodejs-12-rhel7:<v4.5.0>

$ docker pull registry.redhat.io/openshift4/ose-jenkins-agent-maven:<v4.5.0>

$ docker pull registry.redhat.io/openshift4/ose-jenkins-agent-base:<v4.5.0>

To use these images, you can either access them directly from Quay.io or registry.redhat.io or push them into your OpenShift Container Platform container image registry.

Each Jenkins agent container can be configured with the following environment variables.

A JVM is used in all Jenkins agents to host the Jenkins JNLP agent as well as to run any Java applications such as javac, Maven, or Gradle.

By default, the Jenkins JNLP agent JVM uses 50% of the container memory limit for its heap. This value can be modified by the CONTAINER_HEAP_PERCENT environment variable. It can also be capped at an upper limit or overridden entirely.

By default, any other processes run in the Jenkins agent container, such as shell scripts or oc commands run from pipelines, cannot use more than the remaining 50% memory limit without provoking an OOM kill.

By default, each further JVM process that runs in a Jenkins agent container uses up to 25% of the container memory limit for its heap. It might be necessary to tune this limit for many build workloads.

Hosting Gradle builds in the Jenkins agent on OpenShift Container Platform presents additional complications because in addition to the Jenkins JNLP agent and Gradle JVMs, Gradle spawns a third JVM to run tests if they are specified.

The following settings are suggested as a starting point for running Gradle builds in a memory constrained Jenkins agent on OpenShift Container Platform. You can modify these settings as required.

Jenkins agent pods, are deleted by default after the build completes or is stopped. This behavior can be changed by the Kubernetes plugin pod retention setting. Pod retention can be set for all Jenkins builds, with overrides for each pod template. The following behaviors are supported:

You can override pod retention in the pipeline Jenkinsfile:

podTemplate(label: "mypod",
  cloud: "openshift",
  inheritFrom: "maven",
  podRetention: onFailure(), 1
  containers: [
    ...
  ]) {
  node("mypod") {
    ...
  }
}

Source-to-image (S2I) produces ready-to-run images by injecting source code into a container that prepares that source code to be run. It performs the following steps:

Buildah then creates the container image.

Builder images provide their own version of the source-to-image (S2I) scripts that cover the most common use-cases. If these scripts do not fulfill your needs, S2I provides a way of overriding them by adding custom ones in the .s2i/bin directory. However, by doing this, you are completely replacing the standard scripts. In some cases, replacing the scripts is acceptable, but, in other scenarios, you can run a few commands before or after the scripts while retaining the logic of the script provided in the image. To reuse the standard scripts, you can create a wrapper script that runs custom logic and delegates further work to the default scripts in the image.