CPU Load Average Alert

#!/usr/bin/env bash
#
# cpu_load_alert.sh
# Alerts when the system 1-minute load average exceeds a per-core threshold.
# Load is normalised by CPU count so the script behaves the same on any host.
#
set -euo pipefail

# ---- Configuration -----------------------------------------------------------
# Alert when (load1 / cores) is greater than this ratio. 1.00 == fully loaded.
LOAD_RATIO_LIMIT="${LOAD_RATIO_LIMIT:-1.50}"
ALERT_EMAIL="${ALERT_EMAIL:-root@localhost}"
HOSTNAME_FQDN="$(hostname -f 2>/dev/null || hostname)"

# ---- Gather metrics ----------------------------------------------------------
# /proc/loadavg fields: load1 load5 load15 running/total lastpid
read -r load1 load5 load15 _ < /proc/loadavg
cores="$(nproc)"

# Compute the normalised ratio with awk (bash has no floating-point math).
ratio="$(awk -v l="${load1}" -v c="${cores}" 'BEGIN { printf "%.2f", l / c }')"

# Compare the floats; awk returns 1 (true) when over the limit.
over_limit="$(awk -v r="${ratio}" -v lim="${LOAD_RATIO_LIMIT}" \
    'BEGIN { print (r > lim) ? 1 : 0 }')"

# ---- Report ------------------------------------------------------------------
if [[ "${over_limit}" -eq 1 ]]; then
    subject="[CPU ALERT] ${HOSTNAME_FQDN}: load ratio ${ratio} (>${LOAD_RATIO_LIMIT})"
    {
        echo "High CPU load detected on ${HOSTNAME_FQDN}."
        echo
        echo "Cores            : ${cores}"
        echo "Load (1/5/15 min): ${load1} / ${load5} / ${load15}"
        echo "Per-core ratio   : ${ratio}  (limit ${LOAD_RATIO_LIMIT})"
        echo
        echo "Top CPU consumers:"
        ps -eo pid,comm,%cpu --sort=-%cpu | head -n 6
        echo
        echo "Checked at: $(date '+%Y-%m-%d %H:%M:%S %Z')"
    } | mail -s "${subject}" "${ALERT_EMAIL}"
    echo "Alert sent: ratio ${ratio} over ${LOAD_RATIO_LIMIT}."
else
    echo "OK: load ratio ${ratio} within limit ${LOAD_RATIO_LIMIT}."
fi