* Monitor CPU, board, and NVMe temperatures * Monitor NVMe SMART health (wear, errors, status) * Avoid unreliable ACPI thermal zones

—

apt update
apt install lm-sensors smartmontools nvme-cli

—

Path:

/usr/lib/check_mk_agent/plugins/sensors

Content:

#!/bin/bash
echo "<<<sensors>>>"
/usr/bin/sensors

Set permissions:

chmod +x /usr/lib/check_mk_agent/plugins/sensors

—

Path:

/usr/lib/check_mk_agent/local/500_nvme_smart

Content:

#!/bin/bash
# CheckMK local check for NVMe SMART via smartctl
# Version: 0.1
# Scope: /dev/nvme0|9
 
TEMP_WARN=70
TEMP_CRIT=75
USED_WARN=80
USED_CRIT=90
 
for dev in /dev/nvme[0-9]*; do
    [ -c "$dev" ] || continue
 
    disk="$(basename "$dev")"
    out="$(smartctl -a "$dev" 2>/dev/null)"
    [ -n "$out" ] || continue
 
    model="$(printf '%s\n' "$out" | awk -F: '/Model Number/ {sub(/^[ \t]+/, "", $2); print $2; exit}')"
    [ -n "$model" ] || model="unknown"
 
    health="$(printf '%s\n' "$out" | awk -F: '/SMART overall-health self-assessment test result/ {sub(/^[ \t]+/, "", $2); print $2; exit}')"
    critical_warning="$(printf '%s\n' "$out" | awk -F: '/^Critical Warning:/ {sub(/^[ \t]+/, "", $2); print $2; exit}')"
    temp="$(printf '%s\n' "$out" | awk -F: '/^Temperature:/ {sub(/^[ \t]+/, "", $2); print $2; exit}' | awk '{print $1}')"
    used="$(printf '%s\n' "$out" | awk -F: '/^Percentage Used:/ {gsub(/[% ,]/, "", $2); print $2; exit}')"
    spare="$(printf '%s\n' "$out" | awk -F: '/^Available Spare:/ {gsub(/[% ,]/, "", $2); print $2; exit}')"
    spare_thr="$(printf '%s\n' "$out" | awk -F: '/^Available Spare Threshold:/ {gsub(/[% ,]/, "", $2); print $2; exit}')"
    media_err="$(printf '%s\n' "$out" | awk -F: '/^Media and Data Integrity Errors:/ {gsub(/[ ,]/, "", $2); print $2; exit}')"
    err_log="$(printf '%s\n' "$out" | awk -F: '/^Error Information Log Entries:/ {gsub(/[ ,]/, "", $2); print $2; exit}')"
    unsafe="$(printf '%s\n' "$out" | awk -F: '/^Unsafe Shutdowns:/ {gsub(/[ ,]/, "", $2); print $2; exit}')"
    power_hours="$(printf '%s\n' "$out" | awk -F: '/^Power On Hours:/ {gsub(/[ ,]/, "", $2); print $2; exit}')"
 
    [ -n "$health" ] || health="UNKNOWN"
    [ -n "$critical_warning" ] || critical_warning="UNKNOWN"
    [ -n "$temp" ] || temp=0
    [ -n "$used" ] || used=0
    [ -n "$spare" ] || spare=0
    [ -n "$spare_thr" ] || spare_thr=10
    [ -n "$media_err" ] || media_err=0
    [ -n "$err_log" ] || err_log=0
    [ -n "$unsafe" ] || unsafe=0
    [ -n "$power_hours" ] || power_hours=0
 
    state=0
    if [ "$health" != "PASSED" ]; then
        state=2
    elif [ "$critical_warning" != "0x00" ]; then
        state=2
    elif [ "$media_err" -gt 0 ]; then
        state=2
    elif [ "$spare" -le "$spare_thr" ]; then
        state=2
    elif [ "$used" -ge "$USED_CRIT" ]; then
        state=2
    elif [ "$used" -ge "$USED_WARN" ]; then
        state=1
    fi
 
    echo "$state \"NVMe SMART $disk\" used=${used};${USED_WARN};${USED_CRIT}|spare=${spare};;;|media_errors=${media_err};;;|error_log_entries=${err_log};;;|unsafe_shutdowns=${unsafe};;;|power_on_hours=${power_hours};;; $model - health=${health}, critical_warning=${critical_warning}, used=${used}%, spare=${spare}%, media_errors=${media_err}, errlog=${err_log}, unsafe_shutdowns=${unsafe}, power_on_hours=${power_hours}"
 
    tstate=0
    if [ "$temp" -ge "$TEMP_CRIT" ]; then
        tstate=2
    elif [ "$temp" -ge "$TEMP_WARN" ]; then
        tstate=1
    fi
 
    echo "$tstate \"NVMe Temperature $disk\" temp=${temp};${TEMP_WARN};${TEMP_CRIT} $model - ${temp}C"
done

—

check_mk_agent | grep -A5 NVMe

Expected output:

NVMe SMART nvme0n1
NVMe Temperature nvme0n1

—

* Run Service Discovery * Accept the following services:

• NVMe SMART
• NVMe Temperature
• Sensors (CPU, optional board)

—

NVMe temperature thresholds:

* WARN: 70°C * CRIT: 75°C

CPU:

* Use default CheckMK thresholds

—

* NVMe in compact systems (e.g. NUC) tends to run warmer due to limited airflow * Temperatures up to ~65°C under load are typical * Avoid sustained temperatures above 70°C (thermal throttling possible)

—