strix-monitor

KO | EN

gitlite — search

#python #ai #llm #rust #deep-learning #linux #docker #java #awesome #kubernetes #vue #pytorch #security #openai

(★ 13)

Linux retro-themed h/w monitor w/real-time telemetry & historical data. LAN-only by design.

strix-monitor 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

---

A web-based hardware monitor built for Linux systems with AMD GPUs — originally designed around the AMD Strix Halo APU, but works with Nvidia cards too. Real-time telemetry, a retro CRT terminal aesthetic, and built with SSD longevity in mind.

Self-hosted, lightweight, and meant to run as a background service you can pull up from any browser on your LAN.

---

## Features

### Terminal UI
- **CRT aesthetic**: Phosphor glow, scanlines, and flicker effects — all toggleable from the settings panel. Glow color, intensity, and size are individually configurable.
- **Four color palettes**: Green Phosphor, Amber, Ice Blue, and Galaxy Purple. Each has its own tuned glow color, chart fill, and background tones.
- **Custom palette**: Build your own theme from scratch or use any of the four presets as a starting point. Customizes primary/secondary colors, backgrounds, chart fill, and glow.
- **Live font scaling**: Drag a slider in settings to resize the entire UI from 75% to 200% — every element responds instantly via CSS custom properties.
- **Responsive grid**: 12-column layout that keeps widgets aligned across screen sizes.

### Live Monitoring
- **WebSocket streaming**: GPU utilization, clock speed, VRAM usage, unified system memory, CPU per-core load, top processes by RAM, and network I/O — all updating in real time.
- **CPU core heatmap**: Each core rendered as an individual tile. Parked cores show as inactive; active cores glow. Two display modes: rising **Bar** or full-square **Cascade**.
- **Smart memory**: Reads actual system RAM (handles large unified memory architectures like 128 GB) and shows real-time allocation.

### Historical Data View
- Separate history tab with Chart.js line charts for GPU utilization, memory usage, CPU temp/power, and network I/O.
- Selectable time ranges: Last Hour, 6 Hours, 24 Hours, 7 Days, 30 Days, or a custom date range.
- **Hover crosshair** with a pinned readout label that locks to the right edge of each chart.
- **Peak-preserving decimation**: Downsamples thousands of database points to ~2,000 for fast rendering without dropping hardware spikes.
- CSV export for any time range.

### Storage-Conscious Design
- **Buffered writes**: Metrics are held in memory and flushed to SQLite in batches (configurable, default ~60s), keeping disk writes minimal.
- **WAL mode SQLite**: Write-Ahead Logging, delayed sync, and in-memory temp storage reduce write amplification — relevant if you're running this on flash storage 24/7.
- **UTC timestamps throughout**: All records stored in UTC; the history view queries and displays in your local timezone.
- Configurable data retention period (default: 30 days).

---

## Project Structure

```text
strix_monitor/
├── strix_monitor.py      # Flask app, SocketIO events, background metrics loop
├── config.py             # Settings, theme schema, validation
├── database.py           # Thread-safe SQLite, WAL config, query/export
├── system_stats.py       # Hardware collectors (sysfs, psutil, /proc)
├── historical.py         # History view routes and data endpoints
├── templates.py          # Live view HTML template
├── templates/
│   └── historical.html   # History tab template with Chart.js
├── static/
│   ├── terminal.css      # OKLCH design tokens, palette rules, animations
│   └── terminal.js       # Chart.js plugins, WebSocket handling, theme switching
├── themes/
│   └── default_palettes.json  # Preset palette definitions
├── strix-monitor.service # systemd service template
├── verify.py             # Import and syntax sanity check
└── requirements.txt
```

---

## Requirements

- Linux (tested on Ubuntu & CachyOS)
- Python 3.11+
- AMD GPU with ROCm/AMDGPU driver **or** Nvidia GPU with `nvidia-smi`
- A browser on the same machine or LAN

---

## Installation

### 1. Clone

```bash
git clone https://github.com/levanillawafer/strix-monitor.git
cd strix_monitor
```

### 2. Virtual environment

```bash
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
```

### 3. Run

```bash
python strix_monitor.py
```

Open `http://localhost:5000` in your browser, or `http://YOUR_LAN_IP:5000` from another machine on the network.

---

## Running as a systemd Service

If you want it running on boot and restarting on failure, a service template is included.

```bash
# Copy the service file
sudo cp strix-monitor.service /etc/systemd/system/

# Edit it to match your user and paths
sudo nano /etc/systemd/system/strix-monitor.service
```

Update these three lines in the `[Service]` block:

```ini
User=yourusername
WorkingDirectory=/path/to/strix_monitor
ExecStart=/path/to/strix_monitor/.venv/bin/python strix_monitor.py
```

Then enable and start it:

```bash
sudo systemctl daemon-reload
sudo systemctl enable strix-monitor.service
sudo systemctl start strix-monitor.service
```

Check status or follow logs:

```bash
sudo systemctl status strix-monitor.service
sudo journalctl -u strix-monitor.service -f
```

---

## Configuration

Settings are persisted to `~/.config/strix_monitor/config.json` (created automatically on first run). Most settings are available through the in-app Settings panel — palette, glow, scanlines, font scale, core display mode, database write interval, etc.

---

## Security and Deployment Model

Strix.Monitor is designed for use on a **trusted LAN** — the same machine you're monitoring, or another device on your home/office network. It does not implement authentication, CSRF protection, rate limiting, or origin restrictions, and the WebSocket layer accepts connections from any origin.

**Do not expose this service directly to the public internet.** If you need remote access, put it behind a VPN (WireGuard, Tailscale, etc.) or a reverse proxy that handles authentication (nginx + HTTP basic auth, Authelia, etc.).

---

## Vendored Libraries

The following JavaScript libraries are vendored in `static/vendor/` and served locally — there is no CDN dependency at runtime, which keeps the app working on air-gapped networks and removes CDN-compromise as a supply-chain risk:

| Library | Version | License |
| --- | --- | --- |
| [Chart.js](https://www.chartjs.org/) | 4.4.0 | MIT |
| [chartjs-adapter-date-fns](https://github.com/chartjs/chartjs-adapter-date-fns) | 3.0.0 | MIT |
| [Socket.IO Client](https://socket.io/) | 4.5.4 | MIT |

To upgrade a library, download the new release into `static/vendor/`, update the version comment in `vendor_libs.py`, re-run `python3 vendor_libs.py --record` to capture the new hash, and update the `<script>` tag versions in `templates.py` and `templates/historical.html`.

---

## Acknowledgments

Developed iteratively with the help of AI, primarily Claude with additional input from local LLMs running on the same hardware this tool is meant to monitor. Architecture, design decisions, workflows are mine; AI tools accelerated the implementation and helped catch/solve issues through out the process.

---

## License

[MIT](LICENSE) — see the LICENSE file for full text.

strix-monitor

strix-monitor 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

strix-monitor

strix-monitor 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README