Deployment

This documentation provides guidance on deploying the Lomas server using various methods. You can choose the deployment option that best suits your needs:

• Local deployment with devenv: We use devenv to setup our development environment. You can also use it to run a test instance of the Lomas platform on your local machine.

• Onyxia (sspcloud): https://datalab.sspcloud.fr is an instance of the Onyxia datalab platform provided by INSEE (France). It is accessible to national statistical offices as a testing ground for data science projects. If you are part of a national statistical office, you can log in to the sspcloud and start a test instance of the Lomas platform at the click of a button.

• Kubernetes Deployment: We provide a Helm chart to deploy a Lomas instance on a Kubernetes instance. This is the most advanced method of deployment and is best reserved for extensive testing or production purposes.

Server configuration

The Lomas server is configured exclusively via environment variables. This allows to inject sensitive values (e.g. credentials, etc.) via Kubernetes secrets.

Upon startup, the server loads its configuration via a pydantic-settings model defined in server/lomas_server/models/config.py. Environment variables for the server config are prefixed with LOMAS_SERVICE_, you can have a look by running env | grep LOMAS_SERVICE in an active Lomas devenv shell (see here).