#!/bin/bash
# 502 Bad Gateway —— Cloudflare 说源站报错。立刻查 relay。
set -u

echo "════════ 1. relay 还活着吗 ════════"
echo "  服务: $(systemctl is-active lanremote-relay)"
echo "  进程: $(pgrep -c -f 'relay.py' 2>/dev/null || echo 0) 个"
echo "  端口: $(ss -lntp 2>/dev/null | grep -c ':8080')"
echo "  本机 HTTP: $(curl -s -o /dev/null -w '%{http_code}' --max-time 8 http://127.0.0.1:8080/v1/health)"
echo "  官网本机: $(curl -s -o /dev/null -w '%{http_code}' --max-time 8 http://127.0.0.1:8080/)"

echo
echo "════════ 2. relay 日志最后的错误 ════════"
tail -60 /root/lanremote-relay/relay.log 2>/dev/null | tail -30 | sed 's/^/  /'

echo
echo "════════ 3. journal 里的 relay 错误 ════════"
journalctl -u lanremote-relay --since '20 minutes ago' --no-pager 2>/dev/null \
    | grep -iE 'error|traceback|exception|killed|oom|restart' | tail -12 | sed 's/^/  /'
echo "  （没输出=没有明显错误）"

echo
echo "════════ 4. relay 重启过几次 ════════"
journalctl -u lanremote-relay --since '30 minutes ago' --no-pager 2>/dev/null \
    | grep -cE 'Started|Stopped' | sed 's/^/  事件数: /'
journalctl -u lanremote-relay --since '30 minutes ago' --no-pager 2>/dev/null \
    | grep -E 'Started|Stopped|Deactivated|killed' | tail -8 | sed 's/^/    /'

echo
echo "════════ 5. 内存（OOM 会把 relay 干掉）════════"
free -m | sed 's/^/  /'
echo "  relay 内存: $(ps -o rss= -p $(pgrep -f 'relay.py' | head -1) 2>/dev/null | awk '{printf "%d MB", $1/1024}')"
systemctl show lanremote-relay -p MemoryCurrent -p MemoryMax --value 2>/dev/null | sed 's/^/    /'

echo
echo "════════ 6. 隧道连接数 + 落点 ════════"
echo "  连接: $(curl -s --max-time 5 http://127.0.0.1:20241/metrics 2>/dev/null | grep '^cloudflared_tunnel_ha_connections' | awk '{print $2}')"
grep -E 'Registered tunnel connection' /root/lanremote-relay/tunnel.log 2>/dev/null | tail -4 \
    | grep -oE 'location=[a-z0-9]+ protocol=[a-z0-9]+' | sed 's/^/    /'

echo
echo "════════ 7. cloudflared 有没有报回源错误 ════════"
tail -200 /root/lanremote-relay/tunnel.log 2>/dev/null \
    | grep -iE 'origin|502|refused|dial tcp 127|connection refused|no route' | tail -10 | sed 's/^/  /'
echo "  （没输出=cloudflared 没报回源错）"

echo
echo "════════ 8. 公网连打 12 次 ════════"
ok=0
for i in $(seq 1 12); do
    c=$(curl -s -o /dev/null -w '%{http_code}' --max-time 15 \
        -H 'User-Agent: LanRemote/1.4.0' https://relay.xxmxxs.top/)
    [ "$c" = "200" ] && ok=$((ok+1))
    printf "    %2d: %s\n" "$i" "$c"
    sleep 1
done
echo "  可用 $ok/12"

echo
echo "════════ 9. 当前还在跑的长轮询（会不会占满线程）════════"
ss -tn state established '( sport = :8080 )' 2>/dev/null | wc -l | sed 's/^/  到 8080 的连接数: /'
ps -o nlwp= -p $(pgrep -f 'relay.py' | head -1) 2>/dev/null | sed 's/^/  relay 线程数: /'

echo
echo "════════ 10. watchdog 最近 ════════"
tail -10 /root/lanremote-relay/watchdog.log 2>/dev/null | sed 's/^/  /'