#!/bin/bash
# 双协议隧道：同时跑 QUIC 和 HTTP/2 两个 cloudflared 实例。
#
# == 为什么 ==
#
# 实测两种协议各有各的病：
#   QUIC    尾部延迟好（P90 ~2.5 秒），但 DialContext 会
#           "no recent network activity" 超时，建连慢、偶发失败
#    HTTP/2 建连可靠（TCP + MSS 钳制保护），但尾部延迟差（P90 14 秒，
#           TCP 队头阻塞：一条连接上丢一个包，所有流一起卡）
#
# 同一个隧道 ID 可以起**多个副本**，Cloudflare 会把请求分摊到
# 所有副本注册上来的连接上。所以：
#
#     副本 A: --protocol quic  --ha-connections 5
#     副本 B: --protocol http2 --ha-connections 5
#
# 一共 10 条连接、横跨两种协议。QUIC 犯病时 HTTP/2 那半边还在服务，
# 反过来也一样 —— 这是真正的**故障隔离**，而不只是加连接数。
#
# == 资源 ==
#   每个 cloudflared 约占 25 MB。设备 460 MB 总内存、平时可用 300+ MB，
#   两个实例没问题（原来 MemoryMax=160M 是给单实例的，两个各自限 128M）。
set -u

TUNNEL_ID=7fd33d9c-f910-42e3-88aa-752d6b2a9f2f
CONFIG=/root/.cloudflared/config.yml
MAIN=/etc/systemd/system/lanremote-tunnel.service
SECOND=/etc/systemd/system/lanremote-tunnel-h2.service

echo "════════ 1. 备份现有主服务 ════════"
cp -a "$MAIN" "/root/lanremote-tunnel.service.bak.dual.$(date +%s)"
echo "  当前参数: $(grep -oE '\-\-protocol [a-z0-9]+|--ha-connections [0-9]+' "$MAIN" | tr '\n' ' ')"

echo
echo "════════ 2. 主服务改成 QUIC + 5 ════════"
sed -i 's/--protocol [a-z0-9]*/--protocol quic/' "$MAIN"
sed -i 's/--ha-connections [0-9]*/--ha-connections 5/' "$MAIN"
sed -i 's/^MemoryMax=.*/MemoryMax=128M/' "$MAIN"
grep -E 'ha-connections|protocol|MemoryMax' "$MAIN" | sed 's/^/  /'

echo
echo "════════ 3. 创建 HTTP/2 副本服务 ════════"
cat > "$SECOND" <<'UNITEOF'
[Unit]
Description=ykxmz Cloudflare 具名隧道（HTTP/2 副本）
# 和主实例跑同一个隧道 ID —— Cloudflare 会把请求分摊到
# 两个副本注册的全部连接上，形成协议级的故障隔离。
After=network-online.target lanremote-relay.service lanremote-nettune.service
Wants=network-online.target lanremote-relay.service

[Service]
Type=simple
ExecStart=/usr/local/bin/cloudflared tunnel \
    --protocol http2 \
    --edge-ip-version 4 \
    --ha-connections 5 \
    --no-autoupdate \
    --retries 5 \
    --config /root/.cloudflared/config.yml run lanremote
Restart=always
RestartSec=5
MemoryMax=128M
StandardOutput=append:/root/lanremote-relay/tunnel-h2.log
StandardError=append:/root/lanremote-relay/tunnel-h2.log

[Install]
WantedBy=multi-user.target
UNITEOF
echo "  已写入 $SECOND"
cat "$SECOND" | grep -E 'protocol|ha-connections|MemoryMax' | sed 's/^/    /'

echo
echo "════════ 4. 启动，等连接建齐 ════════"
systemctl daemon-reload
systemctl enable lanremote-tunnel-h2 > /dev/null 2>&1
systemctl restart lanremote-tunnel
systemctl restart lanremote-tunnel-h2

for i in $(seq 1 40); do
    sleep 4
    n=$(curl -s --max-time 4 http://127.0.0.1:20241/metrics 2>/dev/null \
        | grep '^cloudflared_tunnel_ha_connections' | awk '{print $2}')
    # 两个实例各自暴露 metrics 端口会冲突，靠下面分别看进程
    [ "${n:-0}" -ge 10 ] && break
done
echo "  metrics 报的连接数: ${n:-?}"

echo
echo "════════ 5. 两个实例的进程和内存 ════════"
for svc in lanremote-tunnel lanremote-tunnel-h2; do
    pid=$(systemctl show $svc -p MainPID --value)
    mem=$(systemctl show $svc -p MemoryCurrent --value)
    proto=$(grep -oE '\-\-protocol [a-z0-9]+' /etc/systemd/system/$svc.service | head -1)
    conns=$(grep -oE '\-\-ha-connections [0-9]+' /etc/systemd/system/$svc.service | head -1)
    printf "  %-26s PID=%-7s 内存=%s MB  %s %s  状态=%s\n" \
        "$svc" "$pid" "$(( ${mem:-0} / 1048576 ))" "$proto" "$conns" \
        "$(systemctl is-active $svc)"
done

echo
echo "  ⚠ 两个实例都想占 20241 metrics 端口，后起的会失败 —— 这是正常的。"
echo "     实际连接数看 cloudflared 日志里的 Registered 行。"
echo "  ---"
echo "  主实例（QUIC）最近注册:"
tail -60 /root/lanremote-relay/tunnel.log 2>/dev/null | grep -c 'Registered tunnel connection' | sed 's/^/    /'
echo "  副本实例（HTTP/2）最近注册:"
tail -60 /root/lanremote-relay/tunnel-h2.log 2>/dev/null | grep -c 'Registered tunnel connection' | sed 's/^/    /'

echo
echo "════════ 6. 等公网可用 ════════"
for i in $(seq 1 25); do
    c=$(curl -s -o /dev/null -w '%{http_code}' --max-time 12 -H 'User-Agent: LanRemote/1.8.2' https://relay.xxmxxs.top/v1/health)
    printf "    第 %d 次: %s\n" "$i" "$c"
    [ "$c" = "200" ] && break
    sleep 3
done

echo
echo "════════ 7. 快速验收（40 次）════════"
ok=0; times=()
for i in $(seq 1 40); do
    out=$(curl -s -o /dev/null -w '%{http_code} %{time_total}' --max-time 25 -H 'User-Agent: LanRemote/1.8.2' https://relay.xxmxxs.top/v1/health)
    c=$(echo "$out" | awk '{print $1}'); ms=$(echo "$out" | awk '{printf "%d", $2*1000}')
    [ "$c" = "200" ] && { ok=$((ok+1)); times+=("$ms"); }
done
stats=$(printf '%s\n' "${times[@]:-0}" | sort -n | awk '{a[NR]=$1} END {printf "中位 %d P90 %d P99 %d 最大 %d", a[int(NR/2)+1], a[int(NR*0.9)+1], a[int(NR*0.99)+1], a[NR]}')
echo "  可用率 $ok/40 = $((ok*100/40))%   $stats ms"