Monitoring & Logs
What to watch
Section titled “What to watch”| Layer | Signals |
|---|---|
| Panel | HTTP errors, login failures, queue depth, DB/Redis health |
| Nodes | Reachability, CPU/memory/disk, agent version drift |
| Servers | Crash loops, backup failures, quota usage |
| Platform | Disk free everywhere, certificate expiry, backup age |
The admin dashboard and system views surface most of this; export it to your existing monitoring (Prometheus/Grafana, Uptime Kuma, plain log shipping) rather than living in the panel UI.
Logs that matter
Section titled “Logs that matter”- Backend container logs — API errors, auth failures, background jobs.
- Agent logs on each node — lifecycle actions, update stages, connectivity.
- Audit log (Admin → Audit Logs) — who changed what in the panel. This is your source of truth for staff actions; protect its retention.
- System errors view — persisted application errors worth triaging regularly.
Alerting minimum
Section titled “Alerting minimum”Page (or at least notify) on: panel down, node unreachable > 5 min, disk > 85%, database backup older than expected, certificate expiring < 14 days. Everything else can be a dashboard. Review and prune alerts quarterly — stale alerts train people to ignore the channel.
When something fires
Section titled “When something fires”Go to troubleshooting with the alert text, the affected host, and a timestamp. Those three facts solve half of all incidents.