Skip to content

RStudio

Rstudio Server can be run as an interactive job and accessed via an SSH tunnel.

Warning

The Spack rstudio package is the Destop Version of rstudio and is NOT for use on the HTCF.

Support

Reminder

The HTCF does not handle Rstudio support requests

After installation, support requests for lab-installed software such as rstudio should be directed to:

  • Google
  • Software support forums or mailing lists
  • The software developers

Note

The texlive package may need modifying before installing rstudio-server. If there is a texlive error during install, the following two lines should be commented out in the texlive package (spack edit texlive):

#    version('live', sha256='74eac0855e1e40c8db4f28b24ef354bd7263c1f76031bdc02b52156b572b7a1d',
#        url='ftp://tug.org/historic/systems/texlive/2021/install-tl-unx.tar.gz')

Building and Installing

Warning

Compiling rstudo requires large amounts of RAM. When setting up the Slurm jobs be sure to include:

--mem-per-cpu=4G --cpus-per-task=<NUM>

in the Slurm parameters. (More than 1 CPU will make the building faster but could cause longer waiting in the queue.)

  1. A custom Spack package needs to be created: spack create rstudio-server

  2. An example rstudio-server package.py (tested against Spack-0.18.0) can be found here. (Older Spack-0.17.2 version here)

Note

Be sure to tell Spack how many CPUs are available:

spack install -j ${SLURM_CPUS_PER_TASK} rstudio-server

Running rstudio-server as a job

A sample rstudio.sbatch script can be found here.

Note

The rstudio.sbatch script will likely need to be modified to add CPU and/or Memory requirements.