#!/bin/bash
# 深挖："被控端一提交配对码，隧道就掉（1033）"
#
# 1033 = cloudflared 没连上 Cloudflare（不是 502 那种源站问题）
# 所以要看：cloudflared 是不是被杀 / OOM / 崩了
set -u

RELAY_DIR=/root/lanremote-relay
URL=https://relay.xxmxxs.top

echo "===== 1. 内存（460MB 的小机器，OOM 是头号嫌疑）====="
free -m | sed 's/^/  /'
echo -n "  swap: "
swapon --show 2>/dev/null | tail -1 || echo "（没有 swap）"

echo
echo "===== 2. 内核有没有杀进程（OOM killer）====="
dmesg 2>/dev/null | grep -iE 'killed process|out of memory|oom' | tail -8 | sed 's/^/  /' || echo "  读不到 dmesg"
echo "  journal 里的 OOM："
journalctl -k --no-pager 2>/dev/null | grep -iE 'killed process|oom' | tail -6 | sed 's/^/    /' || echo "    没有"

echo
echo "===== 3. 服务重启过几次 ====="
for s in lanremote-relay lanremote-tunnel; do
    printf "  %-22s NRestarts=%s  状态=%s\n" "$s" \
        "$(systemctl show $s -p NRestarts --value 2>/dev/null)" \
        "$(systemctl is-active $s 2>/dev/null)"
    echo "    最近 3 条日志："
    journalctl -u "$s" --no-pager -n 3 2>/dev/null | sed 's/^/      /'
done

echo
echo "===== 4. 现在各进程吃多少内存 ====="
ps -eo rss,comm --sort=-rss 2>/dev/null | head -10 | sed 's/^/  /'

echo
echo "===== 5. 隧道连接数 + 公网 ====="
echo -n "  ha_connections = "
curl -s --max-time 5 http://127.0.0.1:20241/metrics 2>/dev/null \
    | grep '^cloudflared_tunnel_ha_connections' | awk '{print $2}'
echo "  公网: HTTP $(timeout 15 curl -4 -s -o /dev/null -w '%{http_code}' \
    -H 'User-Agent: LanRemote/1.4.0' $URL/v1/health 2>/dev/null)"

echo
echo "===== 6. 看门狗最近动作 ====="
tail -10 "$RELAY_DIR/watchdog.log" 2>/dev/null | sed 's/^/  /'

echo
echo "===== 7. 中转自身的内存上限 ====="
echo "  relay 的 MemoryMax: $(systemctl show lanremote-relay -p MemoryMax --value 2>/dev/null)"
echo "  tunnel 的 MemoryMax: $(systemctl show lanremote-tunnel -p MemoryMax --value 2>/dev/null)"
echo "  relay 的 MemoryCurrent: $(systemctl show lanremote-relay -p MemoryCurrent --value 2>/dev/null)"
echo "  tunnel 的 MemoryCurrent: $(systemctl show lanremote-tunnel -p MemoryCurrent --value 2>/dev/null)"

echo
echo "===== 8. 系统整体 OOM 配额 ====="
cat /proc/sys/vm/panic_on_oom 2>/dev/null | sed 's/^/  panic_on_oom=/'
cat /proc/sys/vm/oom_kill_allocating_task 2>/dev/null | sed 's/^/  oom_kill_allocating_task=/'
echo "  设备内存: $(grep MemTotal /proc/meminfo | awk '{print int($2/1024)}') MB"