Deploy to AWS with GitHub Actions and AWS CDK
Updated: Mar 4
This blog explains how you can use AWS CDK to deploy to your AWS account using GitHub Actions CI/CD pipeline.
Feel free to use the code snippets as you please.
Click here and here to view a fully working GitHub AWS Serverless service template that uses the same snippet.
This blog post is available as a video.
Going Over The Job Steps
Let's go over the CI/CD pipeline steps:
AWS account setup
Run linters & unit tests
Run E2E tests
Destroy stack (for dev stack only)
Python & Node Setup
This snippet will setup Python 3.9 and Node v16 in the CI/CD runner.
Setup AWS CDK
It's recommended to use latest pip and AWS CDK version.
In line 4, you should install your Python dependencies with either pip, pipenv, or poetry, depending on your weapon of choice to manage Python dependencies.
These dependencies include development dependencies (pytest, linters, etc.) and service runtime dependencies.
Setup AWS Secrets
You must set up GitHub's repository secrets for this snippet to work.
Under Settings/Secrets/Actions, add 'AWS_SECRET_KEY' and 'AWS_ACCESS_KEY'.
These secrets are used in a predefined IAM role you created for your CI/CD process.
This is a simple example; however, for security reasons, it is best to use an SSO solution (out of the scope of this guide).
The yaml config:
In line 7, choose your AWS region of choice.
Linters & Unit Tests
Right before deployment, it is recommended to run linters such as:
Code complexity checks (radon/xenon)
See the linters example and configuration at my GitHub template here and the makefile that runs the linters commands here.
Once the linters finish, run unit tests as a first service logic gate.
In line3, you must set the correct path to your CDK 'app.py' file.
I usually put all the CDK project files in a 'cdk' folder instead of the root project path.
Once deployment is completed, you should run your E2E tests and make sure your service runs properly on AWS. You can use pytest to create REST API requests or other triggers to test your deployed service.
This step is relevant only for local stack or pull requests where you want to destroy the stack in the end.
As in the deployment stage, make sure set the correct path to your CDK 'app.py' file.