Infrastructure Overview

Multi-provider infrastructure dashboard. Discovers servers from Hetzner Cloud, Hetzner Robot, and Cloudflare, collects metrics via Prometheus, tracks health via blackbox_exporter probes, and shows costs across all resources.

Features

• Multi-provider discovery — Hetzner Cloud, Hetzner Robot (dedicated/auction), Cloudflare (zones, domains, R2, Workers, Pages)
• Prometheus metrics — CPU, RAM, disk, network, containers (via node_exporter + cAdvisor)
• Health monitoring — blackbox_exporter probes (HTTP/TCP/ICMP) configured from the admin UI, with uptime % tracking and incident detection
• Cost tracking — auto-detected from providers + manual cost overrides for auction servers
• Probe targets API — GET /api/probe-targets/{module} serves Prometheus http_sd_configs for blackbox_exporter
• Multi-customer/project — organize servers by customer and project
• Admin UI — manage providers, customers, probe targets, cost overrides, external services
• Optimization suggestions — right-sizing recommendations based on utilization trends

Quick Start

cp .env.example .env
# Edit .env — set DATABASE_URL, OVERVIEW_AUTH_PASSWORD, ENCRYPTION_KEY
docker-compose up -d
# Dashboard: http://localhost:8080

Local Development

python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
# Requires PostgreSQL running
DATABASE_URL=postgresql://overview:overview@localhost:5432/overview python run_local.py

Running Tests

pip install pytest pytest-asyncio respx
# Without PostgreSQL (unit tests only):
python -m pytest tests/ -v
# With PostgreSQL (full suite):
TEST_DATABASE_URL=postgresql://overview:overview@localhost:5432/overview_test python -m pytest tests/ -v

Configuration

All configuration is via environment variables (see .env.example):

Variable	Required	Description
DATABASE_URL	Yes	PostgreSQL connection string
OVERVIEW_AUTH_PASSWORD	Yes (prod)	Dashboard login password
ENCRYPTION_KEY	Yes (prod)	Fernet key for encrypting API keys at rest
POLL_INTERVAL_SECONDS	No	Sync interval (default: 300s)
INFRA_REPO_URL	No	Git repo for hostcfg/WireGuard peer discovery

Providers, customers, and probe targets are managed through the admin UI at /admin/settings.

Health Monitoring Architecture

Admin UI (probe targets)
    ↓
GET /api/probe-targets/{module}  ←  Prometheus (http_sd_configs, every 30s)
    ↓
blackbox_exporter (HTTP/TCP/ICMP probes)
    ↓
Prometheus (probe_success, probe_duration_seconds)
    ↓
ProbeCollector (reads from Prometheus)
    ↓
HealthTracker (history, uptime %, incidents)
    ↓
Dashboard (server page: health checks, uptime, incidents)

Project Structure

app/
├── admin/          Admin UI routes (providers, customers, cost overrides)
├── api/            REST API + probe targets endpoint
├── connectors/     ProbeCollector, Uptime Kuma
├── db/             PostgreSQL async layer (mixins for each domain)
├── health/         HealthTracker (incidents, uptime)
├── metrics/        Prometheus client
├── providers/      Hetzner Cloud, Hetzner Robot, Cloudflare
├── web/            Dashboard routes + Jinja2 templates
├── scheduler.py    Background collection loop
└── reconciler.py   Merges provider data into customer hierarchy

Tech Stack

FastAPI, asyncpg (PostgreSQL), Jinja2, HTMX, httpx, Pydantic