User Guide

Overview

This guide will help you log in to CWRU's High Performance Computing (HPC) cluster and run a test job. Please visit the links - Job Scheduling, Guides and Training for details. 

For new users, it is recommended to get familiarized with the information in HPC Site where you can do keyword search for most of the topics of your interest. If you are seeking for an HPC account, please check the Faculty Guide.

Important Notes

• You need to have an HPC account to log in to the cluster. If you don't have an account, please go through the Access Policy.

• Default Linux file permission is set to 750 but can be changed with explicit request if needed.

• Except Ondemand that just need DUO, you need to use VPN (e.g. Forticlient) outside of the case network.

• There are separate login nodes for research group members, and academic course group members. You will have the same view of your files in all these clusters as they all use network mounted file systems i.e. when you make changes in the file, you are changing the same file irrespective of in which cluster you logged on.

  • pioneer.case.edu                                               # Pioneer Cluster for research

  • markov.case.edu                                               # For Coursework
                                                                                    to get the compute nodes using in batch submission,
                                                                                    please specify "-p markov_cpu" flag i.e. sbatch -p markov_cpu <job-script>

• Don't run the job directly from the login node. You can request the compute node interactively with SLURM "srun" command or submit the job using SLURM script using s"batch" command.  When you are submitting the job using sbatch, make sure that you are doing it from the login node.

• Try to keep a separate directory for each job

  • If everything is on a single directory, the same files might be accessed over and over again at similar timescale, thereby confusing the storage access. It would become worse if the jobs write to the same files at the same time, creating file contention.

• With interactive job submission, you may encounter a termination of connection to HPC after several hours. We recommend you to use batch submission but if you still want to use interactive session ,  please use "ServerAliveInterval flag" as showed:

  • ssh -o ServerAliveInterval=60 <caseID>@pioneer.case.edu

Logging in to the HPC web portal

Web Browser (does not require VPN, but does require Duo dual-factor authentication.)

Pioneer:  https://ondemand-pioneer.case.edu

Markov:  https://ondemand-markov.case.edu 

GUI Client - X2Go

X2Go client : 

• requires a client installation that is available for Windows, Mac, and Linux.

• probably provides the best performance for Graphical Interface applications.

Windows

• Click here to download putty executable or install X2Go (recommended - supported in Linux, Windows, Mac) or MobaXterm (supported in Windows only) for visual access.  Also, consider Windows Subsystem for linux.

• Putty's instructions:

  • Open the executable.

  • In the putty configuration window, enter pioneer.case.edu under "Host Name (or IP address)" 

  • Click open. You will be prompted for a login and password. Enter your CaseID (e.g. xxx125) and Case Password. You may not see the password being entered, that's fine. Just click Enter after you entered your password.

  • You should get a compute node prompt similar to the one shown below. Now you are ready to run your jobs.

[<caseID>@hpc8 ~]$

Mac

• Open the Terminal Application. 

• You will get a prompt similar to the one shown:

bash-3.2$

• At the prompt, type the following and press the Enter key (-X for X-forwarding for Visual access). Make sure that XQuartz is running.

ssh -X <CaseID>@pioneer.case.edu

• Enter your Case Password when prompted.

• You should get a compute node prompt similar to the one shown below. Now you are ready to run your jobs.

[<caseID>@hpc8 ~]$

Linux

Go to Applications->System Tools and choose Terminal

• At the prompt, Type the following and press the Enter key (-X for X-forwarding for Visual).

ssh -X <CaseID>@pioneer.case.edu

• Enter your Case Password when prompted.

• You should get a login node prompt similar to the one shown below. Now you are ready to run your jobs.

[<caseID>@hpc8 ~]

Example: Running MATLAB Application

Request a compute node to run your job by typing the "srun" command at the prompt as shown (--x11 for X-forwarding for visual access). Note that Matlab is memory intensive and it requires more than the default 1gb memory. So, 4gb is requested with --mem flag.  Get more information in Graphical Access.

[<caseID>@hpc8 ~]$ srun --x11 --mem=4gb --pty /bin/bash

You will get the compute node prompt similar to the one below:

[<caseID>@compt335 ~]$

Load the appropriate module. For the MATLAB application, type "module load matlab" at the prompt, as shown below, and press the Enter key:

[<caseID>@compt335 ~]$ module load matlab

Run MATLAB and get the value of pi by typing pi at the prompt and pressing the Enter key (if you do not want GUI, use -nodisplay flag e.g matlab -nodisplay &).

matlab &

>>pi

Congratulations!!! You ran a job in the HPC.

To quit MATLAB:

>>quit

To exit the HPC cluster, type the logout command, as shown below (or press control-d).

[<caseID>@compt335 ~]$ logout