What is measured
System metrics - CPU by core, memory including cache and swap, disk usage and I/O, network throughput per interface - are sampled on an interval and downsampled as they age, so thirty days of history does not become a storage problem.
Alongside them, the state of every service the panel manages: nginx or Apache, PHP-FPM pools, BIND9, Postfix, Dovecot, MariaDB, the SFTP server and each game instance.
Service status
Each service shows running state, uptime, memory use, restart count and the last few lines of its log. The restart count is the underrated one: a service that has restarted forty times today is technically up every time you check on it.
Alerts
- Thresholds on CPU, memory, disk and swap
- Service down or restart-looping
- SSL certificate expiring within a configurable window
- Backup job failed
- Disk quota approaching on a site or game instance
- Failed mail delivery rate
Logs
Logs for managed services are readable from the panel with search and level filtering, and log rotation is configured as part of service setup rather than left to accumulate until a full disk makes it urgent.