Timothy Rudenko is Research Computing’s HPC Systems Administrator. Since Tim spends so much time working on the cluster, he is the perfect person to provide quick and easy tips to help you successfully navigate the cluster. 

“Don’t run jobs on the login node!”

• November 26, 2025

  VSCode sessions are being cancelled by the system on the login nodes on Explorer, please use the Open OnDemand version of VSCode.

• October 29, 2025

  Please be mindful that important job files should not be stored in your /scratch directory.  Research Computing regularly purges /scratch files following our Scratch Space Policy, so make sure that all of your important job files are stored in /project directories.

• September 24, 2025

  Start small and then scale up! This applies to many aspects of using an HPC, but all tests are good tests. If you can test your program or script with a small dataset or problem you will have all the tools you need to scale up with more data and more computational resources (memory, cpu cores, gpu cores). Reach out to the RC team for assistance.

• August 27, 2025

  Waiting for your job to start?

  • Run this command in the terminal: ‘squeue -u username– start’ to see when your job will start
  • If you’re on OOD: Clusters →explorer Shell Access
• July 29, 2025

  Research Computing offers VSCode in Open OnDemand, which allows you to run your VSCode sessions directory on the compute node as SLURM jobs.

• June 26, 2025

  For jobs that don’t need to be run interactively, please use sbatch to submit your jobs to the queue. These jobs will run when resources are available instead of waiting for interactive srun sessions to start.

• May 30, 2025

  To see how busy a partition is, run the command “sinfo -p short”, which will show how many nodes are running jobs (in state “mix” or “alloc”) and how many nodes are in state “idle” (ready and waiting for a job). Or run “squeue -p short | wc -l ”, which will show how many jobs are running on the short partition. Replace with the partition name of interest!

• April 30, 2025

  If you see the error message: “Disk quota exceeded @ dir_s_mkdir – /home/username/…”. It means your home directory has hit the quota limit. No fear, this happens with regular use of the HPC environment. You will need to remove some items in your /home directory. Run this command in the terminal (with your own username) to see which items take up the most space: du -h —max-depth=1 /home/username”

• March 26, 2025

  Use “source activate env_name” to activate conda environments (substitute “env_name” with the name you gave your environment) instead of “conda activate”. The command “conda activate” requires modifying your .bashrc and this can inhibit the performance of OOD applications as well as containerized software.

• February 26, 2025
  • Instead of putting module load or export statements in your .bashrc, put them in a text file called and env.sh, which can be loaded in interactive sessions via “bash env.sh.” Or, you can keep these statements in sbatch scripts.

  • Use a conda environment or python virtual environment to install python packages. Running “!pip install” in a Jupiter Lab session installs the package to a hidden directory in your home (.local) where it can negatively impact the function of future Jupyter Lab notebook sessions.