Creating a CI/CD workflow with GitHub actions and AWS

Automating the process of building applications into docker images and deploying to a cluster.

How do applications that are developed locally run in production?

• They need to be built into images
• The images need to be pushed to an image repository in the cloud
• The images need to be built into a container that can run on a virtual machine or pod in the cloud

What tools can allow this?

• Docker can be used to build a python application into an image (via the steps set out in the Dockerfile)
• AWS has an Elastic Container Registry that can store docker images.
• Images can be built into containers.
• Kubernetes is a container orchestration tool that allows pods to run these images/apps (via containers). The Kubernetes cluster is set up in the cloud, and can be managed by AWS - Elastic Kubernetes Service. 
• AWS also offers Elastic Container Service to run containers in the cloud. 

Using GitHub actions to automate this process

• The steps above can be carried out by writing commands in the terminal locally (e.g. running docker build . builds the app into an image)

• But GitHub actions also allows a virtual machine to run them whenever you push changes to a repo. 

• All of the commands and configuration are written in .yml files in the repo.

• I have created a couple of template Git projects that can automatically execute the process described above (the python application is very simple and just prints a statement). But it illustrates how production grade applications can be managed. 

• Any future projects I work on that need to run in production will be deployed via this process. 

• There is one project template for deploying to Kubernetes via EKS. However, using Kuberntes is slightly overkill for deploying small applications (especially those that just run on a cron schedule), so I have also created a version that deploys the application to ECS. 

• The GitHub READMEs have a lot more information about the steps (see links above).