
All three services are exposed via Traefik's `web` entrypoint. Internal communication uses Kubernetes service names:
- Prometheus: `kube-prometheus-stack-prometheus:9090`
- Alertmanager: `kube-prometheus-stack-alertmanager:9093`
- Grafana: `kube-prometheus-stack-grafana:80`

## Maintenance Notes

- Metrics retention is configured by the Helm chart (default: 15 days)
- Longhorn backups run at **1:00 AM daily** — check Longhorn dashboards for status
- Non-critical service updates at **3:00 AM daily** may cause brief metric gaps
