I've only been learning AWS for a few days and want to practice building a CI/CD pipeline for a small API. To keep the project inexpensive and understand the underlying process, I'm considering running the API in a Docker container on an EC2 instance.
My planned pipeline would run tests and linters, build the Docker image, and push it to a container registry. After a merge, a separate job would connect to the EC2 instance and trigger the deployment.
I'm unsure whether this is a sensible process or follows common practices. I'm also wondering how the deployment job should determine success: should it wait for the new version to start and pass a health check before reporting that the deployment succeeded?
My experience so far is mostly with running tests and linters in GitHub Actions; deployments have previously been handled by a platform or another team. I'd appreciate guidance on this design and on CI/CD concepts in general.
4 Answers
For learning how CI/CD works from first principles, this is a perfectly reasonable exercise. You’ll get to understand image builds, registries, deployment commands, and failure handling. Just recognize that managing an EC2 host yourself means handling more operational work than a managed service would.
Yes, the deployment step should verify the result instead of stopping as soon as a command succeeds. After starting the new container, wait for it to become ready, call a health endpoint or perform a smoke test, and fail or roll back if the check does not pass. That turns deployment success into “the new version is serving correctly,” rather than merely “the remote command ran.”
Choose the highest-level service that satisfies the project. Depending on the API, the options might range from API Gateway and a serverless backend, through Lambda or Fargate, with EC2 as the fallback when you need full host control. The more managed the service, the fewer infrastructure details you need to build and maintain yourself.
If the main goal is learning AWS rather than writing your own deployment machinery, consider using a more managed option. Fargate, for example, can handle container scheduling, health checks, and service rollouts without requiring you to maintain an EC2 instance. EC2 is useful when you specifically need that level of control, but it usually shouldn’t be the default starting point.
Related Questions
Can't Load PhpMyadmin On After Server Update
Redirect www to non-www in Apache Conf
How To Check If Your SSL Cert Is SHA 1
Windows TrackPad Gestures