#!/bin/sh
set -u
LOG_DIR="$HOME/vdi-terminal/logs"
LOG_FILE="$LOG_DIR/overnight-$(date +%Y%m%d).log"
mkdir -p "$LOG_DIR"
log() { printf '%s %s\n' "$(date '+%Y-%m-%d %H:%M:%S %z')" "$*" >> "$LOG_FILE"; }
log "monitor_start pid=$$ interval=60s"
while :; do
    now=$(date '+%Y-%m-%d %H:%M:%S %z')
    shell_pid=$(pgrep -f '^/home/armbian/vdi-terminal/bin/kiosk-shell$' | head -n 1 || true)
    monitor_pid=$(pgrep -f '^/home/armbian/vdi-terminal/bin/vdi-kiosk-monitor' | head -n 1 || true)
    if [ -n "$shell_pid" ]; then
        ps_line=$(ps -o pcpu=,pmem=,rss=,etime= -p "$shell_pid" | tr '\n' ' ')
    else
        ps_line="missing"
    fi
    temp=$(awk '{printf "%.1f", $1/1000}' /sys/class/thermal/thermal_zone0/temp 2>/dev/null || printf "unknown")
    mem=$(awk '/MemTotal:|MemAvailable:|SwapFree:/{printf "%s=%sKB ",$1,$2}' /proc/meminfo 2>/dev/null || true)
    heartbeat="missing"
    if [ -f /tmp/vdi-kiosk.heartbeat ]; then
        heartbeat=$(stat -c '%Y' /tmp/vdi-kiosk.heartbeat 2>/dev/null || printf 'unknown')
    fi
    rdp=$(ss -tnp 2>/dev/null | grep '192.168.1.70:3389' | tr '\n' ';' || true)
    starts=$(grep -c '主进程已启动' /tmp/vdi-kiosk-monitor.log 2>/dev/null || printf 0)
    exits=$(grep -c '主进程已退出' /tmp/vdi-kiosk-monitor.log 2>/dev/null || printf 0)
    crashes=$(grep -ciE 'segmentation fault|bus error|double free|invalid pointer|aborted|terminate called' /tmp/vdi-kiosk-monitor.log 2>/dev/null || printf 0)
    log "sample now=$now shell_pid=${shell_pid:-missing} monitor_pid=${monitor_pid:-missing} ps=[$ps_line] temp_c=$temp heartbeat_epoch=$heartbeat rdp=[$rdp] starts=$starts exits=$exits crashes=$crashes $mem"
    sleep 60
done