# MDF Ops Web UI — Specification

## Overview
A web dashboard for daily server operations. Runs as a Docker container behind Traefik at `cockpit.tekmidian.com`. Wraps the existing `ops` CLI and `offsite.py` module on the Hetzner server.

## Architecture
- **Frontend:** Single-page app (vanilla HTML/CSS/JS or lightweight framework like Alpine.js + Tailwind)
- **Backend:** Python FastAPI app that shells out to `ops` CLI and imports `offsite.py`
- **Deployment:** Docker container, Traefik labels for HTTPS routing
- **Auth:** Simple token-based auth (single admin user, configurable password)

## Pages / Features

### 1. Dashboard (home)
- All containers with status, health, uptime (from `ops status --json`)
- Color-coded: green=healthy, red=unhealthy/exited, yellow=starting
- Group by project (MDF, SeriousLetter, RingsADay, KioskPilot, Coolify)
- Auto-refresh every 30s

### 2. Backups
- **Local backups**: list per project/env with size, age (from `ops backups --json`)
- **Offsite backups**: list per project/env (from `ops offsite list --json`)
- **Backup now** button: triggers `ops backup <project> [env]`
- **Upload to offsite** button: triggers `ops offsite upload <project> [env]`
- **Apply retention** button: triggers `ops offsite retention`
- Show backup age warnings (>24h = yellow, >48h = red)

### 3. Restore
- Select source: local or offsite
- Select project and environment
- If offsite: download first, then restore
- Dry-run option
- Confirmation dialog before executing
- Show progress/output in real-time (SSE or websocket)

### 4. Services
- Per-project service controls: restart container, view logs (last 100 lines)
- Link to Coolify dashboard for deeper management
- Domain links (click to open site)

### 5. System
- Disk usage (from `ops disk`)
- Health check results (from `ops health`)
- Systemd timer status (backup-all, cert-sync)
- Server uptime, load average

## API Endpoints (FastAPI)

```
GET  /api/status                    → ops status --json
GET  /api/backups                   → ops backups --json  
GET  /api/backups/offsite           → ops offsite list --json
POST /api/backup/{project}/{env}    → ops backup project env
POST /api/backup/offsite/upload/{project}/{env}  → ops offsite upload
POST /api/backup/offsite/retention  → ops offsite retention
POST /api/restore/{project}/{env}   → ops restore (with source param)
GET  /api/logs/{project}/{env}/{service}?lines=100  → docker logs
POST /api/restart/{project}/{env}/{service}  → docker restart
GET  /api/system/disk               → ops disk (parsed)
GET  /api/system/health             → ops health (parsed)
GET  /api/system/timers             → systemctl list-timers (parsed)
```

## Server Paths
- ops CLI: `/opt/infrastructure/ops`
- offsite module: `/opt/data/scripts/offsite.py`
- Python: `/opt/data/π/bin/python3`
- registry: `/opt/infrastructure/servers/hetzner-vps/registry.yaml`

## Docker Compose
- Container name: `ops-dashboard`
- Traefik labels for `cockpit.tekmidian.com`
- Mount `/opt/infrastructure`, `/opt/data`, `/var/run/docker.sock` (for docker commands)
- Port: 8080 internal

## Design
- Dark theme, clean, minimal
- Responsive (works on phone too)
- No heavy JS frameworks — keep it light and fast