Node Exporter Down

Summary: A metrics exporter can no longer be scraped.

Consequences: When exporters are no longer scraped, the amount of metrics available to monitoring decrease; this creates a blind spot with respect to failures. Sometimes this alert can indicate the entire host has lost connectivity, as Prometheus exporters communicate via HTTP interfaces.

Possible causes

If the exporter was explicitly removed or its scrape job in Prometheus was renamed, then the alert is expected, and benign. However, there are two other primary causes:

The exporter failed: Check to see if the exporter process (usually a Docker container named *_exporter, but in the case of node_exporter, a systemd service) has failed or is logging errors.

The host has lost connectivity: Check if the host running the exporter has lost connectivity. The hostname label should indicate which host the exporter was running on.