Server

Welcome to the SODAS server documentation.

Before you start using the server, please make sure that you read all the three pages in this part of the documentation:

1. How to access the server
2. How to setup a project and your developer environment
3. How to use the computational resources through Slurm

Please also consider consulting the official UCPH documentation.

We have also collected some general tips & tricks for working with the server and Linux, and some guides for specific tasks.

What's new?

Compared to the old server, there are three main differences:

One: the new server uses Slurm for compute resource management. This means that the server you log in to is just a middle layer, and that to access the compute resources of the old server (i.e. the GPUs, memory, CPU) one must use the Slurm commands (see under usage).

Two: To better structure projects and data, all projects should be stored in the /projects/. These are compute projects and are created through the self service portal (see setup). When you access the compute resources through Slurm, your code and data must be in a projects directory, since that server does not have access to your other network drives.

Three: We strongly recommend that you use uv to manage your Python projects and virtual environments, one reason being the difficulties around the licensing of conda default channels, the other that uv helps you enforce better reproducibility, and is most use cases also faster.

This is also new for us, so please reach out with any questions or suggestions for improvement, but please be patient.