research-dashboard

A tmux-based system monitoring dashboard for ML research workstations. One command gives you GPU, CPU, disk I/O, network, temperatures, RAID status, and live log tailing in a single terminal.

Installation

pip install research-dashboard

Or install from source:

git clone https://github.com/ahbond/research-dashboard.git
cd research-dashboard
pip install -e .

Usage

research-dashboard

Options

Examples

# Launch with defaults
research-dashboard

# Custom session name, skip network pane
research-dashboard --session-name ml-monitor --no-net

# Preview commands without launching
research-dashboard --dry-run

Layout

+---------------------+----------------------+
|    nvtop (GPUs)     |   htop (processes)   |
|                     |                       |
+----------+----------+----------+------------+
| iotop    | nethogs  | sensors  | live logs  |
| (disk)   | (net)    | raid/df  |            |
+----------+----------+----------+------------+

• Top-left -- GPU monitoring via nvtop (falls back to nvidia-smi watch)
• Top-right -- Process monitoring via htop (falls back to btop or top)
• Bottom-left -- Disk I/O via iotop (falls back to iostat)
• Bottom-mid-left -- Network via nethogs (falls back to nload or /proc/net/dev)
• Bottom-mid-right -- Temperatures, GPU stats, RAID status, disk usage
• Bottom-right -- Live log tailing from /tmp/*.log

Requirements

• tmux (required)
• Optional monitoring tools (the dashboard adapts to what is installed):
  • nvtop or nvidia-smi -- GPU monitoring
  • htop, btop, or top -- process monitoring
  • iotop or iostat -- disk I/O
  • nethogs or nload -- network monitoring
  • sensors (lm-sensors) -- temperature readings

Install everything on Ubuntu/Debian:

sudo apt install tmux nvtop htop iotop nethogs lm-sensors

License

MIT