From an Unraid Freeze to a NAS Monitoring and Alerting System

The hardest part of a NAS freeze is not rebooting. It is losing evidence after reboot. Without historical metrics, CPU, memory, disk, Docker, and service failures are only guesses.

Monitoring Topology

flowchart LR
    A["Unraid Host"] --> B["Node Exporter"]
    A --> C["cAdvisor"]
    B --> D["Prometheus"]
    C --> D
    D --> E["Grafana Dashboard"]
    F["HTTP / TCP Services"] --> G["Uptime Kuma"]
    G --> H["Alerts"]

Component Roles

Node Exporter observes host CPU, memory, disk, and network. cAdvisor observes containers. Prometheus scrapes and stores metrics. Grafana visualizes them. Uptime Kuma checks service reachability and alerts.

Debugging Scenario

When load average spikes, check whether CPU is actually saturated, then inspect IO wait, memory pressure, container usage, and disk activity. Host and container metrics must be viewed together.

Checklist

• Collect both host and container metrics.
• Monitor reachability for critical services.
• Organize dashboards around CPU, memory, disk, network, and containers.
• Keep alerts few and actionable.