OnDemand Portal Orientation - Learner Reference

Quick Start Guide

Logging In

1. Go to https://ondemand.hpc.pomona.edu/
2. Enter Pomona username and password
3. Complete DUO MFA
4. Dashboard appears

Main Menu Items

Files - Home Directory: Your /rhome/<myusername> - bigdata: Group research storage - scratch: Temporary fast storage - tmpfs: In-memory storage (for jobs) - Edit Files: Text editor

Jobs - Active Jobs: View running/queued jobs - Job Composer: Submit new jobs - Cluster: View cluster status

Interactive Apps - Jupyter Notebook: Python notebooks with optional GPU - Jupyter Lab: Extended notebook interface - RStudio Server: R statistical computing - Web Terminal: Command-line access

Accounts - Usage: Quota and resource usage - Profile: Account settings

Clusters - Sagehen HPC Shell Access: Web terminal

Common Commands in Web Terminal

File Operations

ls                 # List files
cd dirname         # Change directory
mkdir newdir       # Create directory
cp file1 file2     # Copy file
mv file1 file2     # Move/rename
rm file            # Delete (permanent!)
pwd                # Show current location

Viewing Files

cat filename       # View entire file
head filename      # View first 10 lines
tail filename      # View last 10 lines
less filename      # Page through file
grep pattern file   # Search for pattern

Job Commands

sbatch script.sh   # Submit job
squeue -u $USER    # List your jobs
scancel JOBID      # Cancel job
sinfo              # Cluster status

Module System

module avail       # Available modules
module load name   # Load module
module list        # Loaded modules
module unload name # Remove module

Storage Locations Reference

Location	Size	Purpose	Backed up
/rhome	100 GB	Personal files	Yes
/bigdata	1-10 TB	Research data	Yes
/scratch	Large	Temp files	No
/tmpfs	256 GB	RAM disk	No

Sagehen HPC Cluster Information

Partitions

• normal: General purpose computing (default)
• gpu: GPU-accelerated computing
• memory: High-memory jobs

GPU Types

• A100: 80 GB, for large models
• L40S: 48 GB, for deep learning

Job Limits

• Max time: 720 hours (30 days)
• Max GPUs: 4 per user
• Default memory: See sinfo

Useful Keyboard Shortcuts

In Web Terminal

• Ctrl+C: Stop running command
• Ctrl+L: Clear screen
• Up arrow: Previous command
• Ctrl+R: Search command history
• Ctrl+A: Beginning of line
• Ctrl+E: End of line

In Jupyter Notebook

• Shift+Enter: Run cell
• Ctrl+S: Save notebook
• Esc: Leave edit mode
• A: Add cell above
• B: Add cell below
• D, D: Delete cell

In RStudio

• Ctrl+Enter: Run selected line
• Ctrl+Shift+S: Source file
• Ctrl+Shift+O: Go to file
• Ctrl+Shift+A: Reformat code

Transferring Files

Upload via OnDemand

1. Click file manager
2. Click Upload button
3. Select file(s)
4. Click Start Upload

Download via OnDemand

1. Find file in file manager
2. Right-click → Download
3. Or select and click Download button

From Command Line

scp local_file <myusername>@sagehen.hpc.pomona.edu:/remote/path
scp <myusername>@sagehen.hpc.pomona.edu:/remote/file local_path

OnDemand Features Quick Reference

Dashboard Overview

The OnDemand dashboard is your home page with quick access to:

• Files: Browse and manage all storage locations
• Jobs: Submit and monitor batch jobs
• Interactive Apps: Launch Jupyter, RStudio, terminals
• Account: Check quotas and profile settings
• Help: Click “?” in top navigation for context-sensitive help

File Permissions Quick Guide

chmod 755    Everyone can read; only owner can write
chmod 750    Owner and group can access; others cannot
chmod 700    Only owner can access (most private)
chmod 644    Everyone can read; only owner can write (for files)

Use in web terminal: chmod 755 /path/to/file

Common Errors and Solutions

“File not found” error

• Check file path is correct
• Use absolute path: /rhome/<myusername>/file not ~/file
• Verify file exists in file manager before trying to open
• Check spelling and capitalization

“Permission denied”

• Check file permissions in file manager: Icon with lock means restricted
• Make script executable: chmod +x myscript.sh
• Ask file owner for permission if needed
• Check you’re in the right group

Module not found (in Web Terminal)

• Check correct module name: module avail
• Some modules need a version: module load cuda/12.2.1
• Load the module before running commands: module load miniconda3 then python script.py (there is no standalone python module – Python comes from miniconda3/anaconda3)

Quota exceeded

• Check usage in Accounts → Usage
• Move large files to /bigdata (bigger quota)
• Delete old or temporary files
• Email its-hpc@pomona.edu to request quota increase

Upload is very slow

• Large files take time; be patient
• Faster upload at off-peak hours
• Check internet speed with speedtest.net
• Try uploading smaller chunks if file is huge (>1 GB)

Job won’t start or says “pending”

• Check cluster has available resources: Click Clusters → view sinfo output
• Jobs may wait hours during busy times
• Try requesting fewer resources
• Check job for errors in output file

Can’t launch Jupyter or RStudio

• Long launch times (5-10 min) on busy cluster are normal
• Check OnDemand session hasn’t timed out
• Try stopping the app and launching again
• Check Available resources in Accounts → Usage

OnDemand Best Practices

• Save your work frequently: Sessions timeout after inactivity
• Download notebooks before closing: Jupyter sessions are temporary
• Use /bigdata for large datasets: /rhome has 100 GB limit
• Monitor job output: Check job files for errors after completion
• Log out when done: Closes web terminal and interactive sessions
• Test with small data first: Before processing large datasets

Storage Best Practices

Where to Store	What	Duration	Backed Up
/rhome/<myusername>	Personal files, code	Permanent	Yes
/bigdata/lab	Research data, collaboration	Permanent	Yes
/scratch/job_id	Temporary job files	60 days	No
/tmpfs/job_id	Fast temporary data	During job only	No

Getting Help

• OnDemand Help: Click “?” icon in top right navigation
• Documentation: https://www.pomona.edu/its/
• Email Support: its-hpc@pomona.edu
• In-person Help: HPC office hours (see https://www.pomona.edu/its/)

Tips for Success

1. Start simple: Upload a small file before trying large transfers
2. Use web terminal sparingly: Graphical tools are often easier for file work
3. Save your passwords: Store OnDemand password in password manager
4. Know your storage paths: Learn difference between /rhome, /bigdata, /scratch
5. Check job status regularly: Don’t submit and forget
6. Plan ahead for long jobs: They may queue for hours
7. Keep learning: OnDemand has many features beyond basics

Cheat Sheet: First Time Steps

1. Log into https://ondemand.hpc.pomona.edu/
2. Click Files → Home Directory
3. Upload a test file using Upload button
4. Go to Interactive Apps → Web Terminal
5. In terminal, type: cd /rhome/<myusername> && ls
6. You see your uploaded file
7. Go to Interactive Apps → Jupyter Notebook
8. Click Launch and wait for notebook to start
9. Create a Python cell with: print(“Hello Sagehen!”)
10. Run the cell with Shift+Enter

Congratulations - you’ve used all the main OnDemand features!