Power Sentry Core

PowerSentry Core Skill

Overview

PowerSentry is a distributed infrastructure power monitoring and cost optimization system. It tracks energy consumption and electricity costs for infrastructure servers, featuring real-time monitoring via WebSocket, cost projections, and a centralized web dashboard.

Architecture

┌─────────────────────────────────────────────────────────┐
│                   Control Panel                         │
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐  │
│  │ WebSocket    │  │ HTTP Server  │  │  Database    │  │
│  │   Server     │  │   (Web UI)   │  │  (Central)   │  │
│  │   :8765      │  │   :8080      │  │              │  │
│  └──────────────┘  └──────────────┘  └──────────────┘  │
└─────────────────────────────────────────────────────────┘
                           ▲
                           │ WebSocket
                           │
       ┌───────────────────┼───────────────────┐
       │                   │                   │
┌──────▼────────┐  ┌──────▼────────┐  ┌──────▼─────────┐
│ Worker Agent  │  │ Worker Agent  │  │ Worker Agent N  │
│  (Target)     │  │  (Target)     │  │   (Target)      │
│  Daemon + WS  │  │  Daemon + WS  │  │  Daemon + WS   │
└────────────────┘  └────────────────┘  └────────────────┘

Components

1. Control Panel

• WebSocket Server (:8765): Receives real-time metrics from worker agents
• HTTP Server (:8080): Serves web UI and API endpoints
• Central Database (SQLite): Stores all worker data and historical records
• Web UI (Vue 3 + Tailwind + DaisyUI): Real-time dashboard
• Location: control-panel/server/server.py (Python)

2. Worker Agents

• Daemon (daemon.py): Collects system metrics every minute (CPU, RAM, power)
• WebSocket Client (websocket_client.py): Sends real-time data to control panel
• Local Database (database.py): SQLite backup storage
• CLI (cli.py): Service management (install, start, stop, stats, project)
• Power Model (power.py): Power calculation (CPU TDP, RAM, base power)
• Pricing (pricing.py): French electricity rates (EDF Tarif Bleu, peak/off-peak)

Codebase Structure

proxmox-power-monitor/
├── deploy.sh                    # Deploy worker agent to target (direct or via jump)
├── deploy-control-panel.sh      # Deploy control panel to a server
├── cli.py                       # Worker CLI: install, start, stop, stats, project, cleanup
├── daemon.py                    # Worker daemon: metrics collection loop
├── websocket_client.py          # WebSocket client for worker → control panel
├── database.py                  # SQLite database (local worker storage)
├── power.py                     # Power calculation model
├── pricing.py                   # French electricity pricing
├── metrics.py                   # System metrics collection
├── .env.example                 # Configuration template (copy to .env)
├── src/                         # TypeScript version (legacy/alternative)
│   ├── cli.ts
│   ├── daemon.ts
│   ├── database.ts
│   ├── pricing.ts
│   ├── power.ts
│   └── metrics.ts
├── control-panel/
│   ├── server/server.py         # WebSocket + HTTP server
│   └── ui/index.html            # Web dashboard (Vue 3 + Tailwind + DaisyUI)
├── README.md
├── ARCHITECTURE.md
└── .gitignore

Data Flow

1. Registration: Worker connects and registers with control panel via WebSocket
2. Metrics Collection: Worker collects CPU, RAM, uptime every 60 seconds
3. Real-time Sync: Worker sends metrics via WebSocket to control panel
4. Central Storage: Control panel stores in central database; worker keeps local copy
5. UI Updates: Web dashboard displays live metrics and historical charts
6. Cost Projection: CLI can project costs for any period (days, weeks, months)

Configuration

Worker (.env or env vars)

PROXMOX_HOST=<target-ip>                      # Target server
JUMP_HOST=<jump-ip>                           # Optional: bastion host
JUMP_USER=ubuntu                              # Jump host SSH user
REMOTE_USER=root                              # Target SSH user
DB_PATH=/var/lib/proxmox-power-monitor/data.db
CONTROL_PANEL_URL=ws://<control-panel>:8765/worker

Control Panel (env vars)

PYTHONUNBUFFERED=1

CLI Usage

# Install service
python3 cli.py install
python3 cli.py install --control-panel=ws://<control-panel>:8765/worker

# Manage service
python3 cli.py start
python3 cli.py stop
python3 cli.py status

# View stats
python3 cli.py stats --total
python3 cli.py stats --daily=7
python3 cli.py stats --hourly=24
python3 cli.py stats --recent=60

# Cost projection
python3 cli.py project 14    # Project 14 days

# Cleanup old data
python3 cli.py cleanup --days=90

# Uninstall
python3 cli.py uninstall

WebSocket Protocol

Worker → Control Panel

• register: {"type":"register","worker_id":"...","hostname":"...","version":"1.0.0","capabilities":["metrics","projection"]}
• metrics: {"type":"metrics","worker_id":"...","timestamp":...,"data":{...}}
• heartbeat: Every 30 seconds

Control Panel → Worker

• ack: Acknowledgment of received data
• config: Configuration updates
• reconnect: Reconnection request

Security

• Token-based Authentication: Simple token for WebSocket connections
• TLS Support: Use WSS for production deployments
• Input Validation: All incoming data validated
• Worker Isolation: Each worker can only access its own data
• No secrets in codebase: All IPs/credentials loaded via .env (gitignored)