Deploy Play app on AWS Beanstalk with Docker and Jenkins

It's possible to deploy Play applications on AWS Beanstalk. This project shows how to automatically deploy the latest Play version on AWS Beanstalk as an docker image via Jenkins.

Note: In order to deploy this Play application with automatically with Jenkins you need to fork this project. This is necessary in order to configure the Jenkins service hook for your Github project.

Prerequisites

Dev environment

• OSX 10.10 or above
• Docker
• SBT 0.13.5 or above
• Configured AWS keypair

General

• AWS Account
• Dockerhub account

Enable SBT Native Packager Docker Plugin

The sbt native packager provides experimental docker support. This means we can create a docker files during our Play staging process. Our sample app already uses this feature, take a look at the build.sbt. For more information check out the sbt native packager documentation.

Install additional software on development environment

Before we are using Jenkins to deploy our Play application automatically after each push to Github we are gonna deploy our Play application from our development environment. Do do that we need to install the following tools.

Install EB CLI

The Elastic Beanstalk command line interface helps us to create a Beanstalk application. Follow this guide to install the EB CLI.

Instal AWS CLI

We will use the AWS command line interface to:

• Upload our Play application to an AWS S3 bucket
• Create a new environment version on Beanstalk
• Deploy the Play application to Beanstalk.

Follow this guide to install the AWS CLI.

Configure AWS CLI

Follow this guide to initially configure the AWS CLI.

• AWS Region: Specify the region where your S3 bucket and your Beanstalk instance is located
• Output Format: json

Setup S3 Bucket

A S3 bucket is a good choice to host the different version of our Play application. It is secure, accessible from the dev environment, from Jenkins and Beanstalk.

Our Play application is based on another docker image dockerfile/java. This base image contains Ubuntu and OpenJDK 7 and is hosted on Dockerhub. During the deployment Beanstalk will build and run the docker image. In order to do that it needs to have access to our Dockerhub account. If you have configured your Dockerhub account locally you will find a file called ~/.dockercfg which contains your Dockerhub credentials. Before we can add this file to S3 we need to create a copy with a different name because S3 is not allowing files starting with a dot:

1. On your dev environment create a new copy of your .dockercfg file:

cp ~/.dockercfg ~/dockercfg

1. In your S3 bucket create a folder docker
2. In this folder upload the dockercfg file

Stage app and deploy on Beanstalk (via dev environment)

Create Beanstalk application and environment

First let's create a beanstalk application for our Play app called play-beanstalk:

cd play-beanstalk
eb init

The eb init command will ask you several questions:

1. Select default region: [your AWS region]
2. Select an application to use: Create new Application
3. Enter Application Name: [Your beanstalk application name]
4. Select a platform: Docker
5. Do you want to set up SSH for your instances: y
6. Select a keypair: [Your AWS keypair]

At the end a new Beanstalk application has been created. A beanstalk application is a container which can hold multiple environments. In our case we will create one environment called play-beanstalk-prod:

eb create -i t2.small play-beanstalk-prod

Note: You should use at least a t2.small instance to allocate enough memory for docker and the Play application.

This will create a beanstalk environment with and EC2 instance t2.small, an elastic load balancer and security groups. Also AWS tried to deploy an application directly which will fail. Therefor the Environment health will turn to RED because we haven't specified a Dockerfile and Dockerrun.aws.json file. This is fine for the moment, we will create these files and deploy our first application version later on.

You can leave the AWS print status terminal with CTRL + C after you have seen something like:

WARNING: Environment health has transitioned from YELLOW to RED

Stage Play application + Create docker image

Thanks to the sbt native packager docker plugin we can create docker files during our Play stage process:

sbt clean docker:stage

This creates a target/docker directory containing the Dockerfile based on our build.sbt configuration setting and the docker files itself.

Prepare Docker Image for Beanstalk

Now we need to create a Dockerrun.aws.json file:

cd target/docker
touch Dockerrun.aws.json

This file contains all the necessary information so that Beanstalk can deploy our docker image:

• S3 bucket name which will contain the docker image. Specify the S3 bucket in which you want to host your docker images
• Docker config file which we have added before to S3 that contains our Dockerhub credentials. These are necessary so that Beanstalk can download the dockerfile/java docker image which we are using as a base image in our Play application.
• Container port

Add the following content to this file:

{
 "AWSEBDockerrunVersion": "1",
 "Authentication": {
   "Bucket": "YOUR.S3.BUCKET",
   "Key": "docker/dockercfg"
 },
 "Ports": [
   {
     "ContainerPort": "9000"
   }
 ]
}

During the next steps we will upload our Docker files to the S3 bucket. Beanstalk can either deploy docker images based on a Dockerfile or based on a zip file which contains the Dockerfile. In our case we want to deploy a zip file because we have additionally created a Dockerrun.aws.json file:

zip -r ../../build-play-beanstalk-1.0.zip .
cd ../..

Upload zip to S3 and deploy to Beanstalk

Finally let's upload the created zip file to S3. Then we will create a new Beanstalk application version based and update the Beanstalk environment with this version:

aws s3api put-object --bucket YOUR_S3_BUCKET --key build-play-beanstalk-1.0.zip --body build-play-beanstalk-1.0.zip
aws elasticbeanstalk create-application-version --application-name play-beanstalk --version-label 1.0 --source-bundle S3Bucket=YOUR_S3_BUCKET,S3Key=build-play-beanstalk-1.0.zip

This deploys our Play application to AWS Beanstalk. The app should be now accessible in the internet.

Automate Deployment with Jenkins

We can easily automate this process with Jenkins so that whenever we push our changes to Github, Jenkins will pull this new version and deploy it to Beanstalk automatically.

Prerequisites Jenkins Server

• EC2 Instance with Amazon Linux
• Jenkins with Github Plugin
• Java 7 or above
• AWS CLI

Configure Jenkins project

1. Add your Github credentials as gloal credentials: Jenkins > Credentials > Global credentials > Add Credentials
   • Kind: Username with password
   • Scope: Global
   • Username: [YOUR GITHUB USERNAME]
   • Password: [YOUR GITHUB PASSWORD]
2. Create a new Freestyle project with the name play-beanstalk-prod
3. Select Git in Source Code Management
   • Specify the https repository URL of your Github project, e.g. https://github.com/markusjura/play-beanstalk.git
   • Select your added credentials in Credentials
   • Branches to build: */master
4. Check in Build Triggers the option Build when a change is pushed to GitHub
5. In the Build section add a Execute shell build step to build the Play application and create the docker files:

   # Test and build app
   sbt clean compile docker:stage
   

6. Add another Execute shell build step to add the Dockerrun.aws.json, zip the content, upload the zip to S3 and deploy the application:

   # Docker commands
   # ~~~
   # Retrieve current version from build.sbt
   SBT_APP_VERSION=`less build.sbt | grep version | awk '{print $3}' | sed s/\"//g`
   S3_BUCKET=YOUR_BUCKET_NAME
   APP_VERSION="$SBT_APP_VERSION-$BUILD_NUMBER-$GIT_COMMIT"
   APP_ZIP="build-$JOB_NAME-$APP_VERSION.zip"
   BEANSTALK_APP="play-beanstalk"
   BEANSTALK_ENV="play-beanstalk-prod"
   
   cd target/docker
   
   # Add Dockerrun.aws.json to app
   cp $JENKINS_HOME/beanstalk/Dockerrun.aws.json .
   
   # Zip content based on commit id
   zip -r ../../$APP_ZIP .
   
   # Upload zip to S3 and deploy to Beanstalk
   cd ../..
   
   aws s3api put-object --bucket $S3_BUCKET --key $APP_ZIP --body $APP_ZIP
   
   aws elasticbeanstalk create-application-version --application-name $BEANSTALK_APP --version-label $APP_VERSION --source-bundle S3Bucket=$S3_BUCKET,S3Key=$APP_ZIP
   
   aws elasticbeanstalk update-environment --environment-name $BEANSTALK_ENV --version-label $APP_VERSION
   

7. Save the project

Add Dockerrun.aws.json to Jenkins server

Now add the Dockerrun.aws.json file which have created on our development environment to the EC2 Jekins instance. In our case we add the Dockerrun.aws.json file into $JENKINS_HOME/beanstalk/Dockerrun.aws.json.

Add Jenkins service to Github

Finally Jenkins should deploy the Play application whenever we push something to origin master. Therefor we need to add the Jenkins server as a service to our github project:

1. Go to your Github project, e.g. https://github.com/markusjura/play-beanstalk
2. Navigate to Settings > Webhooks & Services
3. Add a new Jenkins (Git plugin) service
4. Specify your the URL to the Jenkins server, e.g. http://username:[email protected]:8080/github-webhook/

Note: In this simple case is the Jenkins server publicly available. We are transferring the Jenkins username and password directly in the URL.

Note: You need to have access to add the service to your Github project. As a quick test you could fork this project to your Github account.

Test it out

Change some code in the Play application and push the changes to Github origin master. Afterwards the Jenkins project should pick up the changes and depoy the app to AWS beanstalk.