#!/bin/bash
# 撤掉边缘 IP 固定。
#
# 教训（一天内自己推翻自己两次）：
#   第一次：以为优选 IP 能降延迟 -> 错。落点还是 lax，改不了路径。
#   第二次：以为优选 IP 能避开超时 -> 半小时内对，但对不久。
#           当时实测 138ms 的两个 IP，30 分钟后再测全都不通了。
#
#   根因：Cloudflare anycast 的可达性受运营商路由影响，
#   会随时间变化。**固定 IP 等于钉死到一个可能失效的地址上**，
#   比让 GeoDNS 动态解析更糟 —— 后者至少每次重连会重新挑。
#
# 结论：不固定。靠 cloudflared 自己的重试 + --ha-connections 2 冗余。
set -u

echo "===== 1. 清掉 hosts 固定 ====="
before=$(grep -c argotunnel /etc/hosts 2>/dev/null || echo 0)
sed -i '/argotunnel.com/d' /etc/hosts
after=$(grep -c argotunnel /etc/hosts 2>/dev/null || echo 0)
echo "  argotunnel 条目: $before -> $after"

echo
echo "===== 2. 确认解析回到 GeoDNS ====="
python3 - <<'PY'
import socket
for host in ("region1.v2.argotunnel.com", "region2.v2.argotunnel.com"):
    try:
        infos = socket.getaddrinfo(host, 7844, socket.AF_INET)
        ips = sorted({i[4][0] for i in infos})
        print("    %s -> %d 个地址" % (host, len(ips)))
        print("      %s" % " ".join(ips[:5]))
    except OSError as e:
        print("    %s 解析失败 %s" % (host, e))
PY

echo
echo "===== 3. 重启隧道，观察 90 秒连接稳定性 ====="
systemctl restart lanremote-tunnel
stable=1
for i in $(seq 1 9); do
    sleep 10
    conn=$(curl -s --max-time 3 http://127.0.0.1:20241/metrics 2>/dev/null \
           | grep '^cloudflared_tunnel_ha_connections' | awk '{print $2}')
    errs=$(curl -s --max-time 3 http://127.0.0.1:20241/metrics 2>/dev/null \
           | grep '^cloudflared_tunnel_request_errors' | awk '{print $2}')
    echo "    ${i}0s: 连接 $conn  请求错误 $errs"
    [ "$i" -ge 6 ] && [ "$conn" != "2" ] && stable=0
done
[ "$stable" = "1" ] && echo "  ✓ 后段稳定在 2 条连接" || echo "  · 还在收敛"

echo
echo "===== 4. 这段时间的超时告警数 ====="
since=$(date -u -d '100 seconds ago' '+%Y-%m-%dT%H:%M:%S' 2>/dev/null || echo "")
if [ -n "$since" ]; then
    n=$(awk -v s="$since" '$0 >= s' /root/lanremote-relay/tunnel.log 2>/dev/null | grep -c 'i/o timeout')
    echo "  $n 次"
fi

echo
echo "===== 5. 当前落点（应为 2 条）====="
grep -E 'Registered tunnel connection' /root/lanremote-relay/tunnel.log 2>/dev/null | tail -2 \
    | grep -oE 'ip=[0-9.]+ location=[a-z0-9]+ protocol=[a-z0-9]+' | sed 's/^/  /'

echo
echo "===== 6. 公网延迟（15 次，设备侧最坏情况）====="
ok=0; total=0
for i in $(seq 1 15); do
    r=$(curl -s -o /dev/null -w '%{http_code} %{time_total}' --max-time 25 \
        -H 'User-Agent: LanRemote/1.4.0' https://relay.xxmxxs.top/v1/health 2>/dev/null)
    c=$(echo "$r" | cut -d' ' -f1)
    t=$(echo "$r" | cut -d' ' -f2)
    [ "$c" = "200" ] && { ok=$((ok+1)); total=$(echo "$total + $t" | bc); printf "    %2d: %s 秒\n" "$i" "$t"; } \
                     || printf "    %2d: 失败\n" "$i"
    sleep 1
done
avg=$(echo "scale=2; $total / $ok" | bc 2>/dev/null || echo "?")
echo "  可用 $ok/15  平均 ${avg} 秒"