Dashboards

The repository ships Grafana dashboards under grafana/, auto-provisioned by the Docker Compose quickstart against the Prometheus datasource.

Dashboard	File	Focus
iDRAC	grafana/idrac.json	Detailed per-machine view
iDRAC Overview	grafana/idrac_overview.json	Global overview of all machines
Status (alternative)	grafana/status-alternative.json	Detailed per-machine status
PDU	grafana/pdu.json	Rack PDU power, energy, and health

The overview and detail dashboards were contributed by @7840vz.

Previews

Per-machine detail (grafana/idrac.json) — inventory, memory, disks, and NIC link state:

The System template variable picks which machine the detail panels describe:

Fleet overview (grafana/idrac_overview.json) — health rollups, scrape-error counts, and target availability across both export paths:

The same dashboard's sensor row breaks temperature out by inlet, outlet, and CPU:

Alternative per-machine status (grafana/status-alternative.json):

Conventions

• Panels read from the provisioned Prometheus datasource and key off a single system template variable (label_values(idrac_system_machine_info, system)) so one stack can show many machines. system is the canonical host-identity label across both export paths:
• OTLP/snapshot path: the exporter injects it automatically (configurable via otlp.identity_label, default system).
• On-demand scrape-all path: the exporter injects system (and instance) per host on a bare /metrics; Prometheus keeps them with honor_labels: true. For per-target (?target=) scraping, set system via a relabel rule instead — see the Docker Compose quickstart.
• Health metrics map 0 = OK, 1 = Warning, 2 = Critical.
• Any per-second gauge is aggregated with sum/avg in PromQL — never rate() (the exporter already exposes instantaneous values).