Interacting with a major cloud provider may have become a much needed task that's part of your delivery process. GitLab is making this process less painful by providing Docker images that come with the needed libraries and tools pre-installed. By referencing them in your CI/CD pipeline, you'll be able to interact with your chosen cloud provider more easily.
Run AWS commands from GitLab CI/CD
Introduced in GitLab 12.6.
GitLab's AWS Docker image provides the AWS Command Line Interface,
which enables you to run
aws commands. As part of your deployment strategy, you can run
aws commands directly from
.gitlab-ci.yml by specifying GitLab's AWS Docker image.
Some credentials are required to be able to run
Sign up for an AWS account if you don't have one yet.
Log in onto the console and create a new IAM user.
Select your newly created user to access its details. Navigate to Security credentials > Create a new access key.
NOTE: Note: A new Access key ID and Secret access key pair will be generated. Please take a note of them right away.
In your GitLab project, go to Settings > CI / CD. Set the following as environment variables (see table below):
- Access key ID.
- Secret access key.
- Region code. You can check the list of AWS regional endpoints. You might want to check if the AWS service you intend to use is available in the chosen region.
Env. variable name Value
Your Access key ID
Your Secret access key
Your region code
You can now use
awscommands in the
.gitlab-ci.ymlfile of this project:
deploy: stage: deploy image: registry.gitlab.com/gitlab-org/cloud-deploy/aws-base:latest # see the note below script: - aws s3 ... - aws create-deployment ...
NOTE: Note: The image used in the example above (
registry.gitlab.com/gitlab-org/cloud-deploy/aws-base:latest) is hosted on the GitLab Container Registry and is ready to use. Alternatively, replace the image with one hosted on AWS ECR.
Use an AWS Elastic Container Registry (ECR) image in your CI/CD
Instead of referencing an image hosted on the GitLab Registry, you can reference an image hosted on any third-party registry, such as the Amazon Elastic Container Registry (ECR).
To do so, push your image into your ECR
Then reference it in your
.gitlab-ci.yml file and replace the
path to point to your ECR image.
Deploy your application to the AWS Elastic Container Service (ECS)
GitLab provides a series of CI templates that you can include in your project.
To automate deployments of your application to your Amazon Elastic Container Service (AWS ECS)
cluster, you can
AWS/Deploy-ECS.gitlab-ci.yml template in your
GitLab also provides Docker images that can be used in your
gitlab-ci.yml file to simplify working with AWS:
registry.gitlab.com/gitlab-org/cloud-deploy/aws-base:latestto use AWS CLI commands.
registry.gitlab.com/gitlab-org/cloud-deploy/aws-ecs:latestto deploy your application to AWS ECS.
Before getting started with this process, you need a cluster on AWS ECS, as well as related components, like an ECS service, ECS task definition, a database on AWS RDS, etc. Read more about AWS ECS.
The ECS task definition can be:
- An existing task definition in AWS ECS
- A JSON file containing a task definition. Create the JSON file by using the template provided in
the AWS documentation.
Copy the task definition into a new file in your project, for example
<project-root>/ci/aws/task-definition.json. Available in GitLab 13.3 and later.
After you have these prerequisites ready, follow these steps:
Make sure your AWS credentials are set up as environment variables for your project. You can follow the steps above to complete this setup.
Add these variables to your project's
.gitlab-ci.ymlfile, or in the project's CI/CD settings:
CI_AWS_ECS_CLUSTER: The name of the AWS ECS cluster that you're targeting for your deployments.
CI_AWS_ECS_SERVICE: The name of the targeted service tied to your AWS ECS cluster.
CI_AWS_ECS_TASK_DEFINITION: The name of an existing task definition in ECS tied to the service mentioned above.
variables: CI_AWS_ECS_CLUSTER: my-cluster CI_AWS_ECS_SERVICE: my-service CI_AWS_ECS_TASK_DEFINITION: my-task-definition
You can find these names after selecting the targeted cluster on your AWS ECS dashboard:
Alternatively, if you want to use a task definition defined in a JSON file, use
variables: CI_AWS_ECS_CLUSTER: my-cluster CI_AWS_ECS_SERVICE: my-service CI_AWS_ECS_TASK_DEFINITION_FILE: ci/aws/my_task_definition.json
You can create your
CI_AWS_ECS_TASK_DEFINITION_FILEvariable as a file-typed environment variable instead of a regular environment variable. If you choose to do so, set the variable value to be the full contents of the JSON task definition. You can then remove the JSON file from your project.
In both cases, make sure that the value for the
containerDefinitions.nameattribute is the same as the
Container namedefined in your targeted ECS service.
CI_AWS_ECS_TASK_DEFINITION_FILEtakes precedence over
CI_AWS_ECS_TASK_DEFINITIONif both these environment variables are defined within your project.
NOTE: Note: If the name of the task definition you wrote in your JSON file is the same name as an existing task definition on AWS, then a new revision is created for it. Otherwise, a brand new task definition is created, starting at revision 1.
Include this template in
include: - template: AWS/Deploy-ECS.gitlab-ci.yml
AWS/Deploy-ECStemplate ships with GitLab and is available on GitLab.com.
Commit and push your updated
.gitlab-ci.ymlto your project's repository, and you're done!
Your application Docker image will be rebuilt and pushed to the GitLab registry. If your image is located in a private registry, make sure your task definition is configured with a
Then the targeted task definition will be updated with the location of the new Docker image, and a new revision will be created in ECS as result.
Finally, your AWS ECS service will be updated with the new revision of the task definition, making the cluster pull the newest version of your application.
template includes both the
"sub-templates". Do not include these "sub-templates" on their own, and only include the main
AWS/Deploy-ECS.gitlab-ci.yml template. The "sub-templates" are designed to only be
used along with the main template. They may move or change unexpectedly causing your
pipeline to fail if you didn't include the main template. Also, the job names within
these templates may change. Do not override these jobs names in your own pipeline,
as the override will stop working when the name changes.
Alternatively, if you don't wish to use the
to deploy to AWS ECS, you can always use our
aws-base Docker image to run your own AWS CLI commands for ECS.
deploy: stage: deploy image: registry.gitlab.com/gitlab-org/cloud-deploy/aws-base:latest script: - aws ecs register-task-definition ...