Examples · RCC ClusterDocs
Interactive Python, R, Jupyter and Shiny examples
Use these copy-ready examples to run bounded Python, R, Jupyter, and Shiny workflows through Slurm. Each example includes its environment and uses synthetic data so it can be explored safely before adapting it to a project.
Service status: RCC workers and these bounded Slurm development patterns are ready now. Production project vhosts are not yet released.
Included examples
| Directory | Purpose |
|---|---|
examples/interactive-workflows/python |
Python batch analysis with a pinned Conda environment. |
examples/interactive-workflows/r |
R batch analysis with a pinned Conda environment. |
examples/interactive-workflows/jupyter |
JupyterLab inside a Slurm allocation, bound to loopback. |
examples/interactive-workflows/shiny |
Shiny development session inside a Slurm allocation, bound to loopback. |
examples/interactive-workflows/notebooks |
Synthetic Python and R notebooks with sampled distributions, Bray--Curtis PCoA, and waterfall visuals. |
The Python and R batch examples use cpu_short because they finish quickly.
Jupyter and Shiny use bounded interactive allocations and must be stopped when
the attended session ends. Long runners belong in restartable sbatch jobs on
regular compute, not on interactive nodes.
Security boundaries
- Interactive servers bind only to
127.0.0.1. - Users connect through a local SSH tunnel.
- Tokens remain enabled.
- Examples use synthetic data only.
- No example scans hosts, enumerates infrastructure, or exposes internal addresses.
- When released, production web hosting will use the governed vhost process, not a tunnelled Slurm job.
Good cluster patterns
- Use notebooks for inspection and figures.
- Use Slurm scripts for full-scale computation.
- Keep Conda environments and caches on node-local storage when computing.
- Copy final outputs back to the project area.
- Stop interactive jobs when finished.