Running a hosting business without monitoring is guessing. You need to know when servers are overloaded, when disks are filling up, and when customers are having problems, before they tell you.
What to Monitor
Infrastructure Level
- CPU usage: Per-core use, load averages
- Memory: Used, cached, swap usage
- Disk: Space usage, I/O throughput and latency
- Network: Bandwidth usage, packet loss, latency
- Uptime: Service availability checks
Application Level
- Pterodactyl Panel: Web response time, queue processing
- Game servers: Process status, player counts, tick rates
- Database: Query performance, connection counts
- WHMCS: Cron execution, payment processing
Business Level
- Active customers: Signups, cancellations, churn rate
- Support volume: Ticket count, response times
- Revenue: MRR, average revenue per user
Monitoring Tools
Uptime Kuma (Free, Self-Hosted)
Lightweight uptime monitor with a clean dashboard:
- HTTP/HTTPS monitoring
- TCP port checking
- DNS monitoring
- Alert notifications (Discord, Telegram, email)
Installation is simple via Docker:
docker run -d --name uptime-kuma -p 3001:3001 -v uptime-kuma:/app/data louislam/uptime-kuma
Netdata (Free, Self-Hosted)
Real-time system monitoring with beautiful dashboards:
- Per-second metrics for CPU, memory, disk, network
- Automatic anomaly detection
- Zero configuration for basic monitoring
bash <(curl -Ss https://my-netdata.io/kickstart.sh)
Grafana + Prometheus (Free, Self-Hosted)
For advanced monitoring:
- Prometheus collects and stores metrics
- Grafana visualizes them in custom dashboards
- node_exporter provides system metrics
- Custom exporters for application-specific metrics
Alerting
Alert Channels
Set up multiple alert channels:
- Discord webhook: For team notifications
- Email: For business-critical alerts
- SMS/phone: For emergency alerts (server down)
Alert Thresholds
Configure sensible thresholds:
| Metric | Warning | Critical |
|---|---|---|
| CPU usage | >80% for 5min | >95% for 2min |
| Memory | >85% used | >95% used |
| Disk space | >80% full | >90% full |
| Disk I/O | >80% use | >95% use |
| HTTP response | >2 seconds | >5 seconds or down |
Alert Fatigue
Too many alerts is worse than too few: you'll start ignoring them all. Only alert on actionable conditions that need immediate attention.
Automation
Auto-Restart
If a service crashes, restart it automatically:
# Systemd service with auto-restart
[Service]
Restart=always
RestartSec=5
Disk Cleanup
Automate log rotation and temporary file cleanup:
# Cron job for Pterodactyl backup cleanup
0 3 * * * find /var/lib/pterodactyl/backups -mtime +7 -delete
Certificate Renewal
Use Certbot with auto-renewal:
certbot renew --deploy-hook "systemctl reload nginx"
Proactive Maintenance
Weekly Checks
- Review resource usage trends
- Check for pending OS updates
- Verify backup completion
- Review error logs for recurring issues
Monthly Tasks
- Apply security updates
- Test backup restoration
- Review capacity and plan for growth
- Analyze support ticket patterns
Quarterly Reviews
- Performance baseline comparison
- Infrastructure cost review
- Tool and software updates
- Full disaster recovery test
Status Page
A public status page builds customer trust:
- Shows real-time infrastructure status
- Displays planned maintenance windows
- Documents incident history and post-mortems
- Reduces "is it just me?" support tickets
Use Uptime Kuma's built-in status page or a dedicated solution like Instatus.
Monitoring is insurance for your hosting business. The setup takes a few hours, but it prevents the kind of problems that cost days of customers and revenue.