Development

Research code usually needs to be written, tested, and debugged before it can run as a batch job. These tools support that day-to-day development work on the clusters, from interactive notebooks to full-featured editors and command-line helpers.

Warning

Do not run editors, notebooks, or other heavy processes on the login nodes. Use a compute node allocated through Slurm instead.

• Git LFS

  ---

  Install Git LFS on the cluster to track large files in git repositories.

• JupyterHub

  ---

  Start a JupyterLab session on a compute node from a web browser, with no SSH setup required.

• VSCode

  ---

  Edit, run, and debug code on a compute node from a local VSCode window through remote SSH.

• cluv

  ---

  Sync and submit UV-based Python projects across HPC clusters.

• milatools

  ---

  A command-line tool that sets up SSH access to clusters and opens VSCode on a compute node.

Comments