dockerized-flask-on-gce

Bare-bones, Dockerized Flask app running on a Container-Optimized OS Google Compute Engine instance. Infrastructure provisioning and deployment using GitHub Actions.

Local development

Installation

• Create a virtual environment:

$ python3 -m venv .venv

• Activate the virtual environment:

$ source .venv/bin/activate

• Install requirements:

$ pip install -r requirements.txt

Building image

$ docker build -t dockerized-flask-on-gce:local .

Running image

$ docker run -p 6969:6969 dockerized-flask-on-gce:local

Deployment

First time

• Manually create a project on Google Cloud Platform and make a note of the project ID (not the project number)
• Enable the following APIs for the project:
  • Google Compute Engine (GCE)—VM upon which the container will run
  • Identity and Access Management (IAM) API—this enables use of newly created service account
  • Artifact Registry API—private repository for Docker images
  • Cloud Resource Manager API—interacting with containers
• Create a service account with owner access to project
• Create service account key and download
• Manually create a storage bucket to act as the Terraform backend—this must be globally unique and must match the value provided in infra/main.tf

Configuring GitHub Actions

The following Secrets need to be set:

• GCP_PROJECT_ID
• GCP_SERVICE_ACCOUNT_KEY
• GCP_SERVICE_ACCOUNT_KEY_FLATTENED (see below)

Flattening the service account key

Terraform expects a flattened version of the service account key. See these instructions for how to do this.

Business as usual

Commit and push! Successful CI/CD pipelines will result in a new image, tagged with the commit SHA, being built and deployed each time, with old images being pruned from the VM.