Node Exporter exposes operating-system and hardware metrics in Prometheus exposition format. In a Fluent Bit pipeline, node_exporter_metrics polls the compute node, while prometheus_exporter serves the formatted records at an HTTP /metrics endpoint for Prometheus to scrape. section 5.6.1
A minimal integration defines:
- A node_exporter_metrics input with a scrape_interval measured in seconds. section 5.6.1
- A tag and optional tag property for the collected records. section 5.6.1
- A prometheus_exporter output with a listening host, port, and optional Prometheus labels. section 5.6.1
Listing 5.6 Prometheus exporter: prometheus-exporter-output.conf
[SERVICE] flush 5 log_level info [INPUT] name node_exporter_metrics #1 tag node_metrics scrape_interval 5 #2 [OUTPUT] name prometheus_exporter #3 match * port 9885 #4 host 0.0.0.0 add_label fluentbit exporter #5
Use the configuration listing to see the complete five-second polling setup, wildcard output match, port 9885, all-network binding, and fluentbit label. section 5.6.1
Exposed metrics
A scrape can return node_cpu_seconds_total counters labeled by CPU, mode, and exporter identity, alongside disk read and flush counters labeled by device.
