Pi Stats

Overview

Pi Stats is a live Raspberry Pi system monitoring dashboard — Python does the sensing, React does the showing. A zero-dependency Python server (stdlib only) exposes CPU, memory, disk, temperature, network, and process metrics, while a Vite + React dashboard renders them in real time with sparklines, radial gauges, and history charts.

Features

• System Metrics: CPU usage (total + per-core), memory and swap, disk usage and I/O throughput, CPU/GPU temperature, network rx/tx per interface, and top processes — all read straight from /proc, vcgencmd, and ps.
• History: A collector samples every 5 seconds into SQLite (metrics + hourly rollups), with 1H / 24H / 7D time ranges served as ~120 binned points.
• Live Dashboard: Refresh interval selector, metric cards with sparklines and radial gauges, CPU/network/disk-I/O charts, server health panel, per-core grid, top-process tables, and light/dark theming.
• Zero Dependencies (Server): Built entirely on the Python 3 standard library — http.server, sqlite3, and threading. No pip install required.
• Mock Mode: Run the full stack with synthetic, time-varying data for local development or demos — no Raspberry Pi hardware needed.
• Optional Auth: HTTP Basic authentication via a single PI_STATS_PASSWORD environment variable.

Tech Stack

Server:

• Python 3 stdlib only — http.server.ThreadingHTTPServer, sqlite3, threading
• Metrics from /proc/*, statvfs, vcgencmd, and ps
• 5-second collector into SQLite with hourly rollups

Client:

• React 18 + TypeScript, Vite 5
• Tailwind CSS 3, Recharts, lucide-react, shadcn/ui
• Custom useStats hook polling /api/stats with configurable intervals

API

GET /api/stats

Returns a full system snapshot: hostname, uptime, load average, IP, CPU model/cores/frequency and per-core usage, memory and swap, disk usage and I/O, temperatures, aggregate and per-interface network counters, plus top processes by CPU and memory.

GET /api/history?range=1H|24H|7D

Returns ~120 time-binned history points (30s / 12m / 35m bins) covering CPU, memory, temperature, network throughput, disk I/O, disk usage, GPU temp, and load.

All other routes serve the built React client with a path-traversal guard.

Deployment

Designed to run directly on a Raspberry Pi — one Python process serves both the API and the static client build on port 8000. Ships with systemd and crontab @reboot examples for running at boot.