#!/bin/sh
# checks/hw.sh
# Hardware/host metrics: load avg, uptime, free mem, zroot usage, and per
# pool health/capacity plus scrub/error status for every zpool on the box.
set -eu

DIR=$(dirname "$0")
. "$DIR/../lib/common.sh"

LOAD=$(sysctl -n vm.loadavg | tr -d '{}' | sed 's/^ *//;s/ *$//')
json_line "hardware" "load avg" "$LOAD" "ok"

UPTIME=$(uptime | sed 's/.*up //;s/,.*load.*//')
json_line "hardware" "uptime" "$UPTIME" "ok"

MEM_FREE=$(sysctl -n vm.stats.vm.v_free_count)
MEM_PAGE=$(sysctl -n hw.pagesize)
MEM_FREE_MB=$((MEM_FREE * MEM_PAGE / 1024 / 1024))
json_line "hardware" "free mem" "${MEM_FREE_MB} MB" "ok"

DISK=$(df -h /zroot 2>/dev/null | awk 'NR==2{print $5}')
DISK_PCT=${DISK%\%}
if [ -z "$DISK_PCT" ]; then
    DISK_STATUS="warn"
    DISK="unknown"
elif [ "$DISK_PCT" -ge 90 ]; then
    DISK_STATUS="down"
elif [ "$DISK_PCT" -ge 75 ]; then
    DISK_STATUS="warn"
else
    DISK_STATUS="ok"
fi
json_line "hardware" "zroot usage" "$DISK" "$DISK_STATUS"

for pool in $(zpool list -H -o name 2>/dev/null); do
    HEALTH=$(zpool list -H -o health "$pool" 2>/dev/null)
    CAP=$(zpool list -H -o capacity "$pool" 2>/dev/null)
    case "$HEALTH" in
        ONLINE) POOL_STATUS="ok" ;;
        DEGRADED) POOL_STATUS="warn" ;;
        *) POOL_STATUS="down" ;;
    esac
    json_line "hardware" "zpool $pool" "$HEALTH, ${CAP} used" "$POOL_STATUS"

    STATUS_OUT=$(zpool status "$pool" 2>/dev/null) || STATUS_OUT=""
    SCAN=$(printf '%s\n' "$STATUS_OUT" | sed -n 's/^[[:space:]]*scan: //p')
    ERRORS=$(printf '%s\n' "$STATUS_OUT" | sed -n 's/^[[:space:]]*errors: //p')

    case "$ERRORS" in
        "No known data errors") ERR_STATUS="ok" ;;
        "") ERR_STATUS="warn" ;;
        *) ERR_STATUS="down" ;;
    esac
    json_line "hardware" "zpool $pool scrub" "${SCAN:-unknown}" "$ERR_STATUS"
done