AWS CodePipeline Job Worker

Read more about CodePipeline: http://aws.amazon.com/codepipeline/

Build

Dependencies

Install the following tools to build the AWS CodePipeline Job Worker:

• Java SE Development Kit 8
• Apache Maven 3.3

Ant targets

You can use the following ant targets to build the code:

mvn clean    # Cleans the workspace
mvn compile  # Compiles the code
mvn test    # Runs the unit tests
mvn package  # Default target, which compiles the code, runs the unit tests, creates the jar file and deployment bundle

Start

Use the init.d script to start the AWS CodePipeline Job Worker polling for custom actions:

service aws-codepipeline-jobworker start

The script optionally takes the configuration class as a parameter. If you want to run the job worker for third party actions, you can use the following command:

service aws-codepipeline-jobworker start "com.amazonaws.codepipeline.jobworker.configuration.ThirdPartyJobWorkerConfiguration"

You can also specify your own configuration class. It only has to implement the JobWorkerConfiguration interface.

Configuration

The job worker comes with two pre-defined configuration classes: CustomActionJobWorkerConfiguration and ThirdPartyJobWorkerConfiguration. Both inherit from the DefaultJobWorkerConfiguration to share most of the configuration settings.

You can configure the following settings:

// Configure action type id the job worker polls for
public ActionTypeId getActionTypeId() {
    return new ActionTypeId(
        "Deploy",          // Action Type Category: Source, Build, Test, Deploy, Invoke
        "Custom",          // Action Type Owner: Custom or ThirdParty
        "MyCustomAction",  // Action Type Provider: Name of your action type
        "1"                // Action Type Version: e.g. "1"
    );
}

// How frequently the job worker polls for new jobs
private static final long POLL_INTERVAL_MS = 30000L; // e.g. every 30 seconds

// Maximum number of worker threads. Indicates how many jobs can be processed in parallel.
private static final int WORKER_THREADS = 10;

AWS Region

• The AWS region for the custom job worker can be set with the AWS_REGION environment variable, and it will poll for jobs in this region.
• If the environment variable is not set, then the custom job worker will try to use the region of the EC2 instance on which it is running.
• You can also modify getRegion() method in DefaultJobWorkerConfiguration.java.

    private static Region getRegion() {
        return Region.getRegion(Regions.US_EAST_1);
    }

Deployment

The job worker comes with AWS CodeDeploy installation scripts. Set up your application and deployment group in AWS CodeDeploy and run the following command to deploy the agent:

# Replace with your Amazon S3 bucket name
AMAZON_S3_BUCKET=<your-amazon-s3-bucket>
# Make sure that you set up your application and deployment group in AWS CodeDeploy
APPLICATION_NAME=AwsCodePipelineJobWorker
DEPLOYMENT_GROUP_NAME=Production

# Compile the code and create deployment bundle
mvn package

# Install the jar file as a dependency

mvn install:install-file -Dfile=target/jobworker-1.0.jar -DgroupId=com.amazonaws.codepipeline -DartifactId=jobworker -Dversion=1.0 -Dpackaging=jar

# Create a tar archive from the build output
cd build/output
tar -cf ../AwsCodePipelineJobWorker.tar *

# Upload the deployment bundle to your Amazon S3 bucket.
aws s3 cp ../AwsCodePipelineJobWorker.tar s3://$AMAZON_S3_BUCKET/AwsCodePipelineJobWorker.tar

# Start the deployment using AWS CodeDeploy
aws deploy create-deployment --application-name $APPLICATION_NAME --deployment-group-name $DEPLOYMENT_GROUP_NAME --s3-location bucket=$AMAZON_S3_BUCKET,bundleType=tar,key=AwsCodePipelineJobWorker.tar