nada-backend

nada-backend is the API behind the "NAV Data" website.

It serves a GraphQL API for managing dataproducts, and provides functionality for self-service access to the actual data source.

Running locally with tendrils into Google (as a Nada-member)

First ensure that you're authenticated with Google. This assumes that you have gcloud installed, and that Google Cloud Platform has been enabled in myapps.

We require the login to be done in two steps now, apparently.

gcloud auth login --update-adc

Ensure that your .env file is populated with data. Ignore this step if there's already some data that looks right.

This assumes that kubectl is installed and set up with KUBECONFIGS, and that naisdevice is installed and running. This pulls secret data from a Kubernetes cluster and puts it into your .env file.

make env

And ensure that test-sa.json exists and has contents. Ignore this step if the file exists with data.

make test-sa

Start the local database

docker-compose up

Appending a -d flag in the above command detaches the process once the database stuff is running.

Then start the backend!

make local-with-auth

Now you can head over to the nada-frontend repository and learn how to start the frontend.

Deployment

The application needs two GCP service accounts which are mounted in at runtime from two secrets in Google Secret Manager. These are:

• nada-backend-google-credentials
• nada-backend-metabase-sa

It also needs a GCP project where the BigQuery API is enabled to host keys for pseudonymization.

The following values used by the application are also stored in Google Secret Manager in the secret nada-backend:

CLIENT_ID
CLIENT_SECRET
GITHUB_READ_TOKEN
GOOGLE_ADMIN_IMPERSONATION_SUBJECT
METABASE_PASSWORD
METABASE_USERNAME
SLACK_URL