Querying Maximum Temperature Across Linux Server Sensors

Monitoring server temperatures is essential for preventing hardware damage and maintaining system stability. Linux provides several tools to query thermal sensors and identify which components are running hot.

Using lm-sensors

The most reliable approach is lm-sensors, a userspace tool that reads temperature data from kernel drivers:

# Install lm-sensors
sudo apt install lm-sensors    # Debian/Ubuntu
sudo dnf install lm-sensors    # Fedora/RHEL

# Initialize and detect sensors
sudo sensors-detect

# List all sensor readings
sensors

The sensors command displays raw output from all detected thermal zones and their current temperatures.

Extracting the Highest Temperature

To get just the maximum temperature across all sensors:

sensors | grep -oE '[0-9]+\.[0-9]+°C' | sed 's/°C//' | sort -rn | head -1

This pipeline extracts temperature values, removes the degree symbol, sorts numerically in descending order, and returns the highest value.

For a more robust approach using awk:

sensors | awk '/Core|Package|Temp/ {gsub(/[^0-9.]/,"",$NF); print $NF}' | sort -rn | head -1

Using the sysfs Interface

Temperatures are also exposed directly in sysfs without requiring additional tools:

# List all thermal zones
ls /sys/class/thermal/

# Read specific zone temperature (in millidegrees Celsius)
cat /sys/class/thermal/thermal_zone0/temp

# Get highest temperature from all zones
for zone in /sys/class/thermal/thermal_zone*/temp; do
  echo "scale=1; $(cat $zone) / 1000" | bc
done | sort -rn | head -1

Checking hwmon Interface

Many modern sensors report through hwmon:

# Find all hwmon devices
find /sys/class/hwmon -name "temp*_input" | while read f; do
  echo "$(basename $f): $(cat $f)"
done

Divide by 1000 to convert millidegrees to Celsius.

Real-time Monitoring with acpi

For quick checks without installing extra packages:

# View thermal zone information
cat /proc/acpi/thermal_cooling/THM0/temperature

# Or use acpi tool if available
acpi -t

Creating a Monitoring Script

Combine these approaches into a practical script:

#!/bin/bash

# Get max temperature from lm-sensors
get_max_temp() {
  if command -v sensors &>/dev/null; then
    sensors | grep -oE '[0-9]+\.[0-9]+°C' | sed 's/°C//' | sort -rn | head -1
  else
    # Fallback to sysfs
    for zone in /sys/class/thermal/thermal_zone*/temp; do
      echo "scale=1; $(cat $zone) / 1000" | bc
    done | sort -rn | head -1
  fi
}

max_temp=$(get_max_temp)
echo "Maximum temperature: ${max_temp}°C"

# Optional: Alert if exceeds threshold
threshold=85
if (( $(echo "$max_temp > $threshold" | bc -l) )); then
  echo "WARNING: Temperature above ${threshold}°C" >&2
  exit 1
fi

Integration with Monitoring Systems

For continuous monitoring in production environments:

• Prometheus: Use textfile collectors to expose temperature metrics from a cron job
• Telegraf: Built-in inputs.exec plugin can run temperature scripts
• systemd timer: Replace cron with a systemd timer for better integration

Example Prometheus exporter script:

#!/bin/bash
max_temp=$(sensors | grep -oE '[0-9]+\.[0-9]+°C' | sed 's/°C//' | sort -rn | head -1)
echo "node_thermal_max_celsius $max_temp"

Important Considerations

• Permissions: Reading sysfs directly may require root; configure sudo rules if scripting
• Sensor accuracy: Not all sensors are equally reliable; cross-reference multiple readings
• Thresholds: Check your hardware documentation for safe operating temperatures (typically 80-95°C for CPUs)
• Cooling verification: High sustained temperatures indicate cooling issues; check for dust, thermal paste degradation, or fan failures

Always validate temperature readings match your system’s actual load and ambient conditions.