slack-plugin-docker

Configuration and scripts for building and testing slack-plugin (https://github.com/jenkinsci/slack-plugin) with a Docker container. Building with Docker has dubious value, other than providing a consistent environment between developers, because developers are likely to be using an IDE on their host systems that still needs to be setup and configured, but being able to quickly and easily standup a Jenkins server is extremely valuable.

Note that these scripts and configurations were created for a boot2docker/docker-machine environment. If you need to make changes for other environments, make sure to build something in so it keeps working for everyone.

The scripts are pretty simple and are in many cases conveniences, so small personal customizations can often be made by just running the commands directly from the command line.

Option 1: Setup boot2docker

If you need to, install boot2docker (http://boot2docker.io/) and follow the instructions to get it working. You'll basically need the following.

boot2docker init
boot2docker start
eval $(boot2docker shellinit)

Option 2: Setup docker-machine

See https://www.docker.com/docker-machine for instructions on installing and setting up docker-machine. The steps below use a machine name of default, but use whatever settings and naming you prefer. You can even create your machine in a cloud.

docker-machine create default
docker-machine start default
eval $(docker-machine env default)

Building using the Docker container

Create the Docker container for building the slack-plugin project.

Execute the build-docker-image.sh shell script. This creates a docker image tagged as slack-plugin-build, using the Dockerfile file. This image could be uploaded to Docker hub, but it doesn't seem worthwhile right now.

Building the image will probably take quite a while as it needs to download the Maven image from Docker hub, but this only needs to be done once, unless you need to change something in Dockerfile. Subsequent builds will also probably be much faster due to caching.

Note that the Dockerfile creates the expected directories in the container, and copies settings.xml into the root user's Maven home directory (/root/.m2). The local repository in settings.xml is pointed to /var/m2repo, which will be mapped to a directory on the local host when the container is started to build the project. More information about that below.
Build the project using a container by executing the build-plugin.sh shell script.

Note that the shell script assumes that the source code for the slack-plugin project is in a directory called slack-plugin 1 directory above your current directory (i.e. ../slack-plugin).

This is basically what happens when you execute the script.
1. Create a Docker container from the slack-plugin-build image
2. Map the slack-plugin source code directory to a volume on /slack-plugin, so the container can read the project for building, and the target output directory will remain when the container is removed at the end of the build.
3. Map the m2repo in the current directory to a volume on /var/m2repo. This directory will contain the Maven artifacts that are downloaded as dependencies during the build. This means the artifacts are preserved during builds, which is very useful because the largest part of the build is downloading these dependencies. It would be nice to have a fresh start each time, but that would be too time and resource consuming. To get a fresh start, simply clear out the contents of m2repo.
4. Open a psuedo-tty and keep a connection to STDIN open so you can watch the progress.
5. Execute "mvn clean package" to build the project.
6. Stop the container and remove it.

Starting Jenkins

The Jenkins container can be started by executing the start-jenkins.sh shell script. This script is a little smarter than the other scripts, mainly because the Jenkins container runs as a daemon and will continue to run the background until you stop it. The script will detect whether it is already running, whether the container has been started previously and just needs to be restarted, or if it needs to be started for the first time. After startup it should print out the URL for accessing Jenkins from your browser.

Note that for boot2docker/docker-machine it should show the IP address of the VM.

To stop the container: docker stop jenkins

To stop and remove the Jenkins container: reset-jenkins.sh

Debugging

The debug-plugin.sh script can be used for debugging the plugin, after you've built it. There are some issues with debugging through a docker container, so it has limited usefulness. Every time you make code changes, you will need to stop the debug container, rebuild the plugin, and restart the debug container. However, if you do not have a working environment on your local machine, it could still be useful.