Node Exporter
Port: 9100 (internal) Metrics Endpoint: http://109.199.120.120:9100/metrics Node Exporter is a Prometheus exporter that exposes hardware and OS metrics from the host server. It runs as a Docker container with access to the host’s/proc and /sys filesystems, giving Prometheus full visibility into the physical machine.
What It Monitors
| Category | Metrics |
|---|---|
| CPU | Usage per core, idle/user/system/iowait breakdown |
| Memory | Total, available, used, buffers, cached, swap |
| Disk | Free space, used space, read/write I/O, IOPS |
| Network | Bytes in/out, packets, errors, drops per interface |
| System | Load average (1m/5m/15m), uptime, process count |
| File Descriptors | Open FDs, max FDs |
| TCP | Active connections, TIME_WAIT, ESTABLISHED counts |
Grafana Dashboard
The Server Overview dashboard in Grafana displays all Node Exporter metrics with pre-built panels:- CPU Usage % over time
- CPU by mode (user, system, idle, iowait, steal)
- RAM usage (used vs available)
- Disk usage per mount point
- Network traffic (in/out per interface)
- Load average (1m/5m/15m)
- Open file descriptors
- TCP connection states
Key Alert Thresholds
| Metric | Alert Threshold |
|---|---|
| CPU | > 80% sustained for 5 min |
| RAM | > 85% used for 5 min |
| Disk | > 90% used for 5 min |
Why It Matters
Without Node Exporter, you only see container-level metrics. Node Exporter fills the gap by showing:- Whether the host server itself is under pressure
- If disk is filling up due to Docker volumes or logs
- If network bandwidth is saturated
- If there are too many open file descriptors (a sign of connection leaks)