FleetView

A multi-cloud resource dashboard that pulls live server data from AWS, Hetzner, and DigitalOcean into a single, authenticated view — with real-time CPU/RAM/disk metrics powered by Prometheus.

Why this exists

Managing infrastructure across multiple providers means switching between different consoles, each with its own UI, pricing model, and data format. FleetView solves this by normalizing resources from all three providers into a single schema and surfacing them (plus live system metrics) in one dashboard.

Features

• Multi-cloud resource aggregation — fetches EC2 instances (AWS), servers (Hetzner), and droplets (DigitalOcean) concurrently using asyncio.gather, then normalizes them into a common CloudResource model
• JWT authentication — bcrypt-hashed password, signed/expiring tokens, no credentials ever exposed client-side
• Live infrastructure metrics — CPU, RAM, and disk usage pulled from Prometheus (via node_exporter running on the monitored server)
• Single-page dashboard — dark, ops-console-style UI with auto-refresh every 15 seconds
• CI pipeline — GitHub Actions verifies the app imports cleanly on every push
• Containerized — runs anywhere with docker-compose up

Architecture

┌─────────────┐     async/parallel      ┌──────────────┐
│   FastAPI    │ ───────────────────────▶│  AWS EC2 API │
│   Backend    │                          └──────────────┘
│              │ ───────────────────────▶┌──────────────┐
│  (JWT auth)  │                          │ Hetzner API  │
│              │                          └──────────────┘
│              │ ───────────────────────▶┌──────────────┐
│              │                          │DigitalOcean  │
│              │                          │     API      │
└──────┬───────┘                          └──────────────┘
       │
       │ PromQL queries
       ▼
┌──────────────┐      scrapes      ┌──────────────┐
│  Prometheus  │ ◀─────────────────│ node_exporter │
└──────────────┘                    └──────────────┘
       ▲
       │ served to
┌──────┴───────┐
│  Dashboard   │
│ (index.html) │
└──────────────┘

Tech stack

Python · FastAPI · asyncio/httpx · Pydantic · JWT (PyJWT) · bcrypt (passlib) · Prometheus · node_exporter · Docker · Docker Compose · GitHub Actions

Getting started

Prerequisites

• Docker and Docker Compose
• API tokens for the providers you want to monitor (read-only scope is sufficient — see below)
• A Prometheus instance reachable from the backend, with node_exporter running on the server(s) you want metrics for

1. Clone and configure

git clone https://github.com/suatdurkaya/fleetview.git
cd fleetview
cp .env.example .env

Fill in .env:

HETZNER_API_TOKEN=
DIGITALOCEAN_API_TOKEN=
AWS_ACCESS_KEY_ID=
AWS_SECRET_ACCESS_KEY=
AWS_REGION=eu-central-1

ADMIN_USERNAME=
ADMIN_PASSWORD_HASH=
JWT_SECRET_KEY=
JWT_ALGORITHM=HS256

PROMETHEUS_URL=http://localhost:9090

Generate a password hash:

python3 -c "from passlib.context import CryptContext; print(CryptContext(schemes=['bcrypt']).hash('your-password'))"

Generate a JWT secret:

python3 -c "import secrets; print(secrets.token_hex(32))"

Never commit .env. Use read-only API scopes wherever the provider offers them — this app only reads infrastructure state, it never modifies it.

2. Run

docker-compose up --build

Open http://localhost:8000/static/index.html, log in with the password you hashed above.

API reference

All endpoints except /api/login require Authorization: Bearer <token>.

Endpoint	Method	Description
/api/login	POST	Exchange password for a JWT access token
/api/resources	GET	All resources across providers, normalized
/api/metrics	GET	CPU/RAM/disk usage from Prometheus

Full interactive docs at /docs (Swagger UI, auto-generated by FastAPI).

Setting up Prometheus + node_exporter on a monitored server

This app expects a Prometheus instance to already be scraping your servers. On each server you want metrics for:

1. Install and run node_exporter as a systemd service (exposes metrics on :9100)
2. Point your Prometheus scrape_configs at that server's :9100
3. Point PROMETHEUS_URL in .env at your Prometheus instance

Both node_exporter and prometheus should run as dedicated, unprivileged system users (least-privilege principle) — not root.

Roadmap

• "Idle resource" detection — flag servers with sustained low CPU usage
• Multi-server metrics (currently scoped to a single Prometheus target)
• Terraform module to provision the monitoring stack itself
• Kubernetes deployment manifests

Security notes

• API tokens are requested with read-only scope; this app never issues start/stop/restart commands against real infrastructure
• Passwords are bcrypt-hashed, never stored or logged in plaintext
• JWTs expire after 12 hours
• .env is git-ignored and docker-ignored; secrets are injected at runtime, never baked into the Docker image