Continuous Delivery Pipeline for Amazon ECS Using Jenkins, GitHub, and Amazon ECR

This getting started guide is intended to help you set up and configure a continuous delivery pipeline for Amazon EC2 Container Service (Amazon ECS) using Jenkins, GitHub, and the Amazon EC2 Container Registry (Amazon ECR). The pipeline builds Docker images from a GitHub repository, pushes those images to an ECR registry, creates an ECS task definition, and then uses that task definition to create a service on the ECS cluster. We use Jenkins to orchestrate the different steps in the workflow.


To use this guide, you must have the following software components installed:

Step 1: Build an ECS Cluster#

  1. Create an AWS access key and a secret key by opening a terminal window, and then typing the following:

    aws iam create-access-key --user-name <user_name>

    <user_name> is an IAM user with Adminstrator Access.


    AdministratorAccess is a managed policy that allows attached entities to perform all actions against all resources. Although we’re using it here for convenience, you should remove the AdministratorAccess policy from your IAM user when it’s no longer needed.

  2. Copy the output from the previous command to a text file.

  3. Create an AWS profile on your local machine. At a command prompt, type the following:

    aws configure

    At the prompts, paste your AWS access key ID and AWS secret key ID, enter the preferred region (us-west-2), and then choose json as the output format.

  4. Create an SSH key in the us-west-2 region. You will use this SSH key to log in to the Jenkins server to retrieve the administrator password.

    1. Open the EC2 console.

    2. Under Networking & Security, choose Key Pairs.

    3. Choose Create Key Pair.

    4. For Key pair name, type a name for the key pair, and then choose Create.

      A file is downloaded to your default download directory.

    5. (OS X only) Change the working directory to your download directory and change permission so only the current logged-in user can read it. <file_name> is the name of the .pem file you downloaded:

      chmod <file_name> 400
  5. Clone the GitHub repository that contains the AWS CloudFormation templates to create the infrastructure you will use to build your pipeline.

    1. Open a command prompt and clone the GitHub repository that has the template:

      git clone
    2. Change the working directory to the one that was created when you cloned the repository. At the command prompt, type or paste the following. <key_name> is the name of an SSH key in the region where you're creating the ECS cluster:

      aws cloudformation create-stack --template-body file://ecs-cluster.template --stack-name EcsClusterStack --capabilities CAPABILITY_IAM --tags Key=Name, Value=ECS --region us-west-2 --parameters ParameterKey=KeyName,ParameterValue=<key_name> ParameterKey=EcsCluster,ParameterValue=getting-started ParameterKey=AsgMaxSize,ParameterValue=2


      Do not proceed to the next step until the Stack Status shows CREATE_COMPLETE. To get the status of the stack type aws cloudformation describe-stacks --stack-name EcsClusterStack --query 'Stacks[*].[StackId, StackStatus]' at a command prompt.

Step 2: Create a Jenkins Server#

Jenkins is a popular server for implementing continuous integration and continuous delivery pipelines. In this example, you'll use Jenkins to build a Docker image from a Dockerfile, push that image to the Amazon ECR registry that you created earlier, and create a task definition for your container. Finally, you'll deploy and update a service running on your ECS cluster.

  1. Change the current working directory to the root of the cloned repository, and then execute the following command:

    aws cloudformation create-stack --template-body file://ecs-jenkins-demo.template --stack-name JenkinsStack --capabilities CAPABILITY_IAM --tags Key=Name,Value=Jenkins --region us-west-2 --parameters ParameterKey=EcsStackName,ParameterValue=EcsClusterStack


Do not proceed to the next step until the Stack Status shows CREATE_COMPLETE.To get the status of the stack type aws cloudformation describe-stacks --stack-name JenkinsStack --query 'Stacks[*].[StackId, StackStatus]' at a command prompt.
  1. Retrieve the public host name of the Jenkins server. Open a terminal window and type the following command:

    aws ec2 describe-instances --filters "Name=tag-value", "Values=JenkinsStack" | jq.Reservations[].Instances[].PublicDnsName
  2. Copy the public host name.

  3. SSH into the instance, and then copy the temp password from /var/lib/jenkins/secrets/initialAdminPassword.

    1. On OS X, use the following command:

      ssh -i <full_path_to_key_file> ec2-user@<public_hostname>

      For Windows instructions, see Connecting to Your Linux Instance from Windows Using PuTTY.

    2. Run the following command:

      sudo cat /var/lib/jenkins/secrets/initialAdminPassword
    3. Copy the output and log out of the instance by typing the following command:


Step 3: Create an ECR Registry#

Amazon ECR is a private Docker container registry that you'll use to store your container images. For this example, we'll create a repository named hello-world in the us-west-2 (Oregon) region.

  1. Create an Amazon ECR registry by running the following command:

    aws ecr create-repository --repository-name hello-world --region us-west-2
  2. Record the value of the URL of this repository because you will need it later.

  3. Verify that you can log in to the repository you created (optional).

    Because the Docker CLI doesn't support the standard AWS authentication methods, you need to authenticate the Docker client in another way so Amazon ECR knows who is trying to push an image. Using the AWS CLI, you generate an authorization token that you pass into the Docker login command.

    • If you’re using OS X, type: $(aws ecr get-login)
    • If you’re running Windows, type: aws ecr get-login | cmd


    This command will not succeed unless you have the Docker client tools installed on your machine and the Docker Virtual Machine is running. The output should say login succeeded.

Step 4: Configure Jenkins First Run#

  1. Paste the public host name of the Jenkins server from step 2.3 into a browser.

  2. Paste the password you copied from the /var/lib/jenkins/secrets directory from Step 2: Create a Jenkins Server (step 2.4) in the password field, and then choose Next.

  3. Choose Install suggested plugins.

  4. Create your first admin user by providing the following information:

    • Username: username
    • Password: password
    • Confirm password: password
    • Full name: full_name
    • Email address: email_address
  5. Choose Save and finish.

  6. Choose Start Using Jenkins.

  7. Install the Jenkins plugins.

    In this step, you install the Amazon ECR plugin and the Cloudbees Docker build and publish plugin. You use the Amazon ECR plugin to push Docker images to an Amazon ECR repository. You use the Cloudbees Docker build and publish plugin to build Docker images.

    1. Log in to Jenkins with your user name and password.
    2. On the main dashboard, choose Manage Jenkins.
    3. Choose the Manage plugins tab.
    4. Choose the Available tab.
    5. Select the Cloudbees Docker build and publish plugin and the Amazon ECR plugin.
    6. Choose Download now and install after restart.
    7. Choose Restart Jenkins when installation is complete and no jobs are running.

Step 5: Create and Import SSH Keys for GitHub#

In this step, you create an SSH key and import it into GitHub so you can log in to GitHub over SSH.

  1. If you’re running OS X, open a terminal window. If you’re running Windows, open a Git Bash shell. Run the following command:

    ssh-keygen -t rsa -b 4069 -C
  2. Accept the file location and type a passphrase.

  3. Ensure ssh-agent is enabled by running the following command:

    eval "$(ssh-agent -s)"
  4. Add the SSH key to the agent:

    ssh-add ~/.ssh/id_rsa


    If you already have a key named id_rsa, choose another name.

  5. Copy the contents of the file to the clipboard. On OS X you can use the following command:

    pbcopy < ~/.ssh/
  6. Log in to GitHub. (If you don't have a GitHub account, follow the instructions at to sign up for one.)

    1. In the top-right corner of any page, choose your profile picture, and then choose Settings.
    2. In the user settings sidebar, choose SSH and GPG keys.
    3. Choose New SSH key or Add SSH key.
    4. Type a title for the key.
    5. Paste your key in the key field.
    6. Click Add SSH key.
    7. If prompted, confirm your GitHub password.

Step 6: Create a GitHub Repository#

In this step, you create a repository to store your Docker file and its dependencies.

  1. Create a repository.
    1. Log in to Github.
    2. Choose Start a project or New repository.
    3. Type a name for the repository.
    4. Choose Create repository.
  1. Push code to your repository.

    1. Open a terminal window (OS X) or Git Bash shell (Windows).

    2. Change the working directory to the root of the hello-world repository you cloned earlier.

    3. Delete the hidden .git directory.

      If you’re running OS X, type rm -fR .git. Otherwise, type del /S /F /Q .git.

    4. Reinitialize the repository and push the contents to your new GitHub repository using SSH by running the following command:

      git init
    5. Stage your files:

      git add .
      git commit -m "First commit"
    6. Set your remote origin.

      If you are using SSH, run the following command:

      git remote add origin  '<your_repo>.git'

      If you are using HTTPS, run the following command:

      git remote add origin '<your_repo>.git'


      If you created the SSH key for GitHub on your machine, you can use either method. The HTTPS method requires that you enter your GitHub user name and password at the prompts.

    7. Push your code to GitHub by running the following command:

      git push -u origin master

      This project includes a file named taskdef.json. You can view it in the GitHub interface or with a text editor on your local machine. This file is the JSON representation of your ECS task definition.


      You must supply values for the family and name keys. These are used later in the Jenkins execution scripts. You have to set the value of the image key to %REPOSITORY_URI%:v_%BUILD_NUMBER%. You will use this mechanism to add the Jenkins build number as a tag to the Docker image.

  2. Enable webhooks on your repository so Jenkins is notified when files are pushed.

    1. Browse to your GitHub repository.

    2. Choose Settings.

    3. Choose Integrations & Services.

    4. Choose Add service.

    5. In the search field, type Jenkins (github plugin).

    6. Enter the public FQDN/github-webhook/ of your Jenkins server in the Jenkins URL field, prepended by your Jenkins user name and password.


      If your Jenkins password contains special characters, you have to encode them using URL escape codes.

      Be sure you have a trailing slash (/) at the end of the URL. For example, http://username:password@FQDN/github-webhook/.

    7. Choose Update service.

Step 7: Configure Jenkins#

In this step you will create a Jenkins Freestyle project to automate the tasks in your pipeline.

  1. Create a freestyle project in Jenkins.

    1. Log in to Jenkins
    2. Choose New Item, and then type a name for the project.


    Make sure the name does not include spaces.

    1. Choose freestyle project from the list of project types.

    2. Choose OK.

    3. Under the source code management heading, choose the git button.

    4. In the repository URL field, type the name of your GitHub repository, e.g.,<repo>.git.

    5. In Credentials, choose the GitHub credentials you created in step 1 of this procedure.

    6. Under build triggers, choose Build when a change is pushed to GitHub.

    7. Scroll to the build section, and then choose Add a build step.

    8. Choose Execute shell.

    9. In the command field, type or paste the following text:

      DOCKER_LOGIN=`aws ecr get-login --region us-west-2`
    10. Choose Add a build step, and then choose Docker Build and Publish.

    11. In the repository name field, type the name of your ECR repository.

    12. In the tag field, enter v_$BUILD_NUMBER.

    13. In Docker registry URL, type the URL of your Docker registry. Use only the fully qualified domain name (FQDN) of the ECR repository you created earlier in Step 3: Create an ECR Registry.

    14. Click Add a build step.

    15. Choose execute shell.

    16. In the command field, type or paste the following text. Be sure to replace <ECR_repo> and <cluster_name> with the appropriate values from your environment:

      FAMILY=`sed -n 's/.*"family": "\(.*\)",/\1/p' taskdef.json`
      NAME=`sed -n 's/.*"name": "\(.*\)",/\1/p' taskdef.json`
      #Store the repositoryUri as a variable
      REPOSITORY_URI=`aws ecr describe-repositories --repository-names ${REPOSITORY_NAME} --region ${REGION} | jq .repositories[].repositoryUri | tr -d '"'`
      #Replace the build number and respository URI placeholders with the constants above
      sed -e "s;%BUILD_NUMBER%;${BUILD_NUMBER};g" -e "s;%REPOSITORY_URI%;${REPOSITORY_URI};g" taskdef.json > ${NAME}-v_${BUILD_NUMBER}.json
      #Register the task definition in the repository
      aws ecs register-task-definition --family ${FAMILY} --cli-input-json file://${WORKSPACE}/${NAME}-v_${BUILD_NUMBER}.json --region ${REGION}
      SERVICES=`aws ecs describe-services --services ${SERVICE_NAME} --cluster ${CLUSTER} --region ${REGION} | jq .failures[]`
      #Get latest revision
      REVISION=`aws ecs describe-task-definition --task-definition ${NAME} --region ${REGION} | jq .taskDefinition.revision`
      #Create or update service
      if [ "$SERVICES" == "" ]; then
        echo "entered existing service"
        DESIRED_COUNT=`aws ecs describe-services --services ${SERVICE_NAME} --cluster ${CLUSTER} --region ${REGION} | jq .services[].desiredCount`
        if [ ${DESIRED_COUNT} = "0" ]; then
        aws ecs update-service --cluster ${CLUSTER} --region ${REGION} --service ${SERVICE_NAME} --task-definition ${FAMILY}:${REVISION} --desired-count ${DESIRED_COUNT}
        echo "entered new service"
        aws ecs create-service --service-name ${SERVICE_NAME} --desired-count 1 --task-definition ${FAMILY} --cluster ${CLUSTER} --region ${REGION}


      Before saving this project, be sure that the variable CLUSTER is set to the name you gave your cluster, the REPOSITORY_NAME is set to the name of your ECR registry, and the REGION is set to the region where you created your ECS cluster.

    17. Click Save.

  2. Make a change to a file in your repository, for example,, and push it to GitHub (from Step 6: Create a GitHub Repository, repeat step 2, or use the GitHub interface). If you configured things correctly, Jenkins pulls the code from your Git repository into a workspace, builds the container image, pushes the container image to ECR, creates a task and service definition, and starts your service.

  1. Confirm that your service is running.

    1. Log in to the AWS Management Console

    2. Under Compute, choose EC2 Container Service.

    3. Choose the name of the cluster you created earlier. For example, getting-started.

    4. On the Services tab, choose the name of the service you created. For example, hello-world-service.

    5. On the Task tab, choose the RUNNING task.

    6. Under Containers, click the twisty next to the container name.

    7. Under Network bindings, choose the IP address in the External Link column.

      You should see the following image in your browser:

      DockerCloudHelloWorld image