A terminal user interface (TUI) for monitoring Slurm clusters. Built with Textual for DGX H100 clusters.

⚠️ Development Notice: This project is mainly implemented by LLM (Sonnet 4/GPT-4) and is not complete, has bugs. Contributions are welcome, including major changes.

• Job monitoring with live updates
• Node status display with GPU availability
• GPU count display with partition info
• Script viewer with syntax highlighting
• Output tracking (stdout/stderr)
• Search and filtering
• Tabbed TUI interface
• Keyboard shortcuts
• gpustat-web integration for real-time GPU monitoring

# 1. Use uvx
$ uvx --from git+https://github.com/MilkClouds/smon.git smon

# 2. Use uv tool
$ uv tool install git+https://github.com/MilkClouds/smon.git
$ smon

$ pip install git+https://github.com/MilkClouds/smon.git
$ smon

smon

smon --help                    # Show help
smon --refresh 10              # Set refresh interval to 10 seconds
smon --user alice              # Filter jobs by user
smon --partition gpu           # Filter jobs by partition
smon --gpustat-web URL         # Enable gpustat-web integration

• Job information: JobID, User, State, Partition, Resources
• GPU/CPU/memory usage and timing
• Select job to view details, script, and output

• Shows script for selected job
• Bash syntax highlighting
• Modal view with s key

• stdout/stderr for selected jobs
• Real-time refresh toggle (t)
• Manual refresh (Ctrl+R)

• Node status and availability
• GPU/CPU/memory per node
• gpustat-web integration (side-by-side view)

smon can display real-time GPU status from gpustat-web alongside the Slurm node information.

1. Make sure gpustat-web is running on your cluster (e.g., http://10.50.0.111:48109/)

2. Run smon with the --gpustat-web option:

   smon --gpustat-web http://10.50.0.111:48109/

3. Or add to config file (~/.config/smon/config.json):

   {
     "gpustat_web_url": "http://10.50.0.111:48109/"
   }

The Nodes tab will show the Slurm node table on the left and live GPU status from gpustat-web on the right.

• Python ≥ 3.11
• Slurm cluster with squeue, sinfo, and scontrol commands
• Terminal with color support

• Textual - TUI framework
• Rich - Text formatting

• https://github.com/wjwei-handsome/Slurmer

This TUI project is primarily implemented using LLM assistance (Sonnet 4/GPT-4) and is incomplete with known bugs. Contributions are welcome:

• Bug fixes
• Feature improvements
• Code refactoring
• Documentation
• Major changes
• Testing

Feel free to open issues or submit pull requests.