LM Sensors

linux hardware monitoring

Useful for monitoring system statistics such as CPU core temperatures, fan speeds, and voltages.

Setup

After installing the lm_sensors package, run sensors-detect as root to walk through loading any kernel modules necessary to read sensor data:

1
# sensors-detect

This will probe for hardware monitoring chips and offer to add the detected modules to your system's module loading configuration. Once done, run sensors to see the output:

1
$ sensors

Watching Values

For continuous monitoring, use the -u flag for raw values or watch for changes:

1
$ watch -n 1 sensors

Configuration

Sensor output can be customized through /etc/sensors3.conf or drop-in files under /etc/sensors.d/. This is useful for:

Renaming sensor labels to something meaningful
Setting warning and critical thresholds
Applying correction factors for sensors that read slightly off
Hiding sensors that aren't useful

Example configuration to customize a sensor:

1
2
3
4
chip "coretemp-isa-0000"
  label temp2 "Core 0"
  set temp2_max 85
  set temp2_crit 95

Integration with Other Tools

The sensor data exposed by lm_sensors is readable by most monitoring systems (Prometheus node_exporter, collectd, Nagios plugins, etc.) through the sysfs interface at /sys/class/hwmon/. The sensors command is just a nice frontend for reading the same data.