This is the footprint we can use to start any service based on python from. It's optional but will save us a lot of time.
- This template is created as a template. Selected this template when creating a new repository.
- Edit the .env file and adjust variables accordingly
- Create a Python environment locally and install only the necessary libraries to keep the image as small as possible.
- Review Dockerfile to select the most appropriate image. Now we are using Python:3 (pretty slim!)
- Use make to work with this repository:
In order to run this service locally, first authenticate in gcloud with the appropriate service account credentials, install DVC and download the latest model version.
pip install 'dvc[gs]'
dvc pull -r model-tracker-gcp
make build
It builds the image in the local docker repository.
make run
It runs the application in the container. It's configured for hot reloading. Everytime you make a change in the code the app will refresh.
make logs
Outputs in the terminal the output of the application
make stop
Stop the application
make bash
Enter the command line of the image for debugging. Shouldn't need to do this very often
make auth
Authenticates gcloud and docker
make deploy
Deploys the application in the appropriate GCP Artifact repository. By defualt the repository should be named as the image created locally. IF it's not Make file should be modified.