I was recently asked to setup Jenkins on our CI Box that is running CentOS 6. I wasn’t able to find a good guide, so I decided to create my own.

Installation

The first thing to do is to install development tools. Not all of these are needed, but they will save headaches in the future.

$ yum groupinstall 'Development Tools'

Next, the default version of Java included in CentOS is not compatible with Jenkins. You will need to remove the current version, and install Java 1.6.

$ yum remove java
$ yum install java-1.6.0-openjdk

The Jenkins wiki doesn’t mention this, but I had trouble running Jenkins without Tomcat 6.

$ yum install tomcat6
$ wget -O /etc/yum.repos.d/jenkins.repo http://pkg.jenkins-ci.org/redhat/jenkins.repo
$ rpm --import http://pkg.jenkins-ci.org/redhat/jenkins-ci.org.key
$ yum install jenkins

Start up Jenkins as a service: $ sudo service jenkins start

Now Jenkins should be up and running, navigate to http://localhost:8080 in order to verify that it is working.

Useful Plugins

  • GitHub Plugin - Integration between Jenkins and GitHub.
  • Rake Plugin - Allows Jenkins to invoke Rake tasks as build steps.
  • Ruby Plugin - This plugin will let users use Ruby in the build scripts.
  • RVM Plugin - Runs your jobs in the RVM managed ruby+gemset of your choice.
  • Green Balls - Changes Jenkins to use green balls instead of blue for successful builds.

GitHub Integration

In order to get Jenkins to be able to pull a private repo from GitHub, you will need to generate private and public keys. I found the easiest way to do this is as the jenkins user.

$ su -l -p jenkins
$ ssh-keygen -t rsa -C "[email protected]"

Now upload the public key to Github, and test to verify it worked:

$ ssh -T [email protected]

Configuring a Ruby/Coffeescript Project

  1. Create a new project, and give it a name.
  2. Under teh Source Code Management setting, enter in it’s repository. I use the SSH git repo.
  3. Under Build Triggers, if you want to poll your repository every 5 minutes, enter: 5 * * * *
  4. Set the RVM enviroment: [email protected]
  5. Configure the build.

Configuring the build is a customized process. I will include my build scripts so that you can base yours off of them.

# Setup
export QMAKE=/usr/bin/qmake-qt47
rm -rf ${WORKSPACE}/db/development.sqlite3
rm -rf ${WORKSPACE}/db/test.sqlite3
rm -rf ${WORKSPACE}/db/production.sqlite3
ln -fs /home/USER/Documents/PROJECT/development.sqlite3 ${WORKSPACE}/db/development.sqlite3
ln -fs /home/USER/Documents/PROJECT/development.sqlite3 ${WORKSPACE}/db/test.sqlite3
ln -fs /home/USER/Documents/PROJECT/development.sqlite3 ${WORKSPACE}/db/production.sqlite3
bundle install

#Run Jasmine and Rspec
bundle exec jasmine init
bundle exec rake jasmine:headless
bundle exec rspec

#Setup Cucumber and Run
/etc/init.d/xvfb start
export DISPLAY=:99.0 && bundle exec cucumber

Note: If you plan on running Cucumber tests, you need to install xvfb and modify your init.d.

$ yum install xvfb
#!/bin/bash
#chkconfig: 345 95 50
#description: Starts xvfb on display 99
if [ -z "$1" ]; then
echo "`basename $0` {start|stop}"
	exit
fi

case "$1" in
start)
	/usr/bin/Xvfb :99 -screen 0 1280x1024x24 &
;;

stop)
	killall Xvfb
;;
esac 

Wrap Up

My personal recommendation is to have multiple builds in Jenkins, instead of one large build. We currently have our jasmine/rspec suite checking for new commits every 5 minutes, while our cucumber build checks for new commits every hour. This helps to provide quick feedback to the developers, and prevents them from having to wait for Cucumber to finish running.

If you use this guide and run into any issues, leave a comment and I’ll try to keep this guide updated. I wrote this guide about a week after I setup Jenkins, so it is possible I left out some steps.