What it is

Node exporter is a Prometheus exporter for hardware and OS metrics exposed by *NIX kernels. It is written in Go and ships with pluggable metric collectors, so the set of measurements it gathers can be tuned per host rather than fixed at build time. The project lives in the Prometheus ecosystem, is published under the Apache-2.0 license, and has been developed for roughly thirteen years, with the repository showing 13,782 stars and 2,708 forks at the time of writing.

The concrete problem it solves is the gap between a running machine and a monitoring system that needs numbers about that machine. Prometheus scrapes HTTP endpoints, but kernels do not expose CPU, memory, filesystem, disk, and network state in a scrape-ready format. Node exporter bridges that gap by reading kernel and system interfaces and republishing them as Prometheus metrics on a single HTTP endpoint, listening on port 9100 by default. It is the standard way to bring host-level telemetry into a Prometheus deployment without writing custom collection code for each operating system.

Key capabilities

• Exposes hardware and OS metrics from *NIX kernels over HTTP for Prometheus scraping.
• Provides pluggable collectors, enabled with --collector. flags and disabled with --no-collector. flags.
• Supports --collector.disable-defaults to start from an empty collector set and enable only what is needed.
• Offers include and exclude flags on selected collectors, such as --collector.filesystem.mount-points-exclude, to filter by device or mount point pattern.
• Covers collectors including arp, cpu, diskstats, ethtool, and filesystem, with support varying by operating system.
• Runs as a single Go binary with no runtime dependency on an external database or message broker.

Who uses it and how

• Operators monitoring bare-metal and virtual Linux hosts alongside a Prometheus server that scrapes port 9100.
• Enterprise Linux and Fedora administrators installing the packaged build golang-github-prometheus-node-exporter from EPEL 7, 8, or 9.
• Teams automating rollout with the Prometheus Community Ansible role.
• Container-based deployments running quay.io/prometheus/node-exporter:latest with host networking, host PID, and a read-only bind mount of / to /host, passing --path.rootfs=/host so the exporter reads the host filesystem rather than the container's.
• Hosts where the timex collector is required, which needs the additional --cap-add=SYS_TIME Docker flag for the underlying syscalls.

Getting started

Install from distribution packages, the Prometheus Community Ansible role, or the container image quay.io/prometheus/node-exporter:latest. Run the binary or container and point Prometheus at the default HTTP port 9100; the project links a step-by-step guide on prometheus.io for newcomers.

When to use

docker run -d \
  --net="host" \
  --pid="host" \
  -v "/:/host:ro,rslave" \
  quay.io/prometheus/node-exporter:latest \
  --path.rootfs=/host

---
version: '3.8'

services:
  node_exporter:
    image: quay.io/prometheus/node-exporter:latest
    container_name: node_exporter
    command:
      - '--path.rootfs=/host'
    network_mode: host
    pid: host
    restart: unless-stopped
    volumes:
      - '/:/host:ro,rslave'

--collector.filesystem.mount-points-exclude=^/(dev|proc|sys|var/lib/docker/.+|var/lib/kubelet/.+)($|/)