Prometheus / node_exporter
Read node_exporter metrics through the Prometheus HTTP API.
NasDash queries the Prometheus server, normally on port 9090. Prometheus must already scrape node_exporter on the monitored machine. Do not enter the exporter’s address or its /metrics endpoint as the server address.
In NasDash
- Open Settings → Integrations → Monitoring → Prometheus.
- Enter host
prometheus.example.comand Port9090. Usehttps://on the host for HTTPS. - Leave credentials empty if none are required. For Basic authentication, fill username and password; for bearer authentication, leave username empty and put the token in the password field.
- Test and list machines, select the node_exporter instance and save. The target must exactly match its
instancelabel, for example192.0.2.10:9100. Manage it in Settings → Widgets → Devices.
Readings and diagnostics
The collector uses /api/v1/query for CPU, memory and filesystem usage. CPU temperature, cores, uptime, load and network rates appear when the corresponding metrics exist. Virtual interfaces and temporary filesystems are filtered. GPU readings are not collected by this integration.
Discovery reads node_uname_info, falling back to up{job=~".*node.*"}. If the instance is missing or has no data, check the scrape target and exact label in Prometheus. CPU and network rates need samples over a two-minute window. See monitoring and troubleshooting.