Link Conda to RStudio

Accessing RStudio on Rockfish Using R from a Conda Environment

This tutorial guides you through launching RStudio Server on Rockfish using R installed inside a Conda environment.

Connect to Rockfish (Login Node)

Open a terminal and run:

ssh YourUserID@login.rockfish.jhu.edu

Enter your Rockfish password when prompted.

Create a Module for Your Conda Environment

1. Load Anaconda:

   module load anaconda3/2024.02-1
   

2. Activate your Conda environment:

   conda activate my_conda_env
   

3. Generate a modulefile for your environment:

   conda_to_lua.sh
   

   This will create a module named after your Conda environment (e.g., my_conda_env) in your user module path.

4. Deactivate the Conda environment:

   conda deactivate
   

Create a SLURM Script with r-studio-server.sh

r-studio-server.sh -c 10 -t 02:00:00 -p shared

Argument Explanation:

• -c: Number of cores (each core provides ~4 GB RAM)

• -t: Runtime in HH:MM:SS

• -p: Partition - express: up to 4 cores / 8 hours - shared: up to 32 cores / 36 hours - parallel: up to 48 cores / 72 hours

You can run r-studio-server.sh --help to view all available options.

Edit the Generated SLURM Script

nano R-Studio-Server.slurm.script

Inside the file:

1. Locate the line:

   module restore
   

2. Directly below it, add a line to load your Conda module:

   module load own my_conda_env
   

3. Find the line:

   export R_LIBS_USER=${HOME}/R/4.2.1
   

4. Replace it with the path to your Conda environment’s R library:

   export R_LIBS_USER=/home/YOUR_USERNAME/.conda/envs/my_conda_env/lib/R/library
   

Replace YOUR_USERNAME with your Rockfish username and my_conda_env with your Conda environment name.

Save and exit:

• Ctrl + O, then Enter to save

• Ctrl + X to exit

Update RStudio Configuration

nano .r-studio-variables

Replace the following lines:

Line 13:

module load gfbf/2023b RStudio-Server/2023.12.1+402-gfbf-2023b-Java-11-R-4.4.1

Line 14:

ml -R/4.4.1-gfbf-2023b

Save and exit:

• Ctrl + O, then Enter to save

• Ctrl + X to exit

Submit the Job

sbatch R-Studio-Server.slurm.script

This will generate a file named:

rstudio-server.job.<jobid>.out

Locate the Output File

ls -ltr rstudio-server.job*

View Connection Instructions

cat rstudio-server.job.<jobid>.out

Inside the file, you’ll find two important things:

• The SSH tunnel command

• The localhost URL to open in your browser

Set Up the SSH Tunnel (Local Machine)

On your local terminal (not logged into Rockfish), run:

ssh -N -L <PORT>:<node>:<PORT> YourUserID@login.rockfish.jhu.edu

Replace values accordingly. You’ll be prompted for your password — enter it and leave this terminal open.

Access RStudio

From the .out file, copy the URL starting with:

http://localhost:<PORT>/...

Paste it into your web browser and log in with your Rockfish credentials.