#!/bin/bash
# 双协议隧道的长时稳定性测量。
#
# 和之前的 longstab.sh 区别：现在有**两个** cloudflared 实例，
# 只有先起的那个能绑上 20241 metrics 端口，所以不能只看 metrics
# 的连接数 —— 要分别从两个日志里数"当前活跃连接"。
#
# 每轮记录：
#   * QUIC 实例 / HTTP2 实例 各自最近注册的连接数
#   * 公网可用率
#   * 掉线、dial 失败、530/502 事件
set -u

BASE=https://relay.xxmxxs.top
UA='LanRemote/1.8.2'
MINUTES=${1:-30}
LABEL=${2:-dual}
ROUNDS=$((MINUTES * 6))

OUT=/root/longstab_${LABEL}.txt
MARK_Q=$(wc -l < /root/lanremote-relay/tunnel.log 2>/dev/null || echo 0)
MARK_H=$(wc -l < /root/lanremote-relay/tunnel-h2.log 2>/dev/null || echo 0)

echo "════════ 双协议隧道长测 · $LABEL · ${MINUTES} 分钟 ════════" | tee "$OUT"
echo "  QUIC:   $(grep -oE '\-\-protocol [a-z0-9]+|--ha-connections [0-9]+' /etc/systemd/system/lanremote-tunnel.service | tr '\n' ' ')" | tee -a "$OUT"
echo "  HTTP2:  $(grep -oE '\-\-protocol [a-z0-9]+|--ha-connections [0-9]+' /etc/systemd/system/lanremote-tunnel-h2.service | tr '\n' ' ')" | tee -a "$OUT"
echo "  MTU: $(cat /sys/class/net/wlan0/mtu)" | tee -a "$OUT"
echo "  开始: $(date '+%H:%M:%S')" | tee -a "$OUT"

okA=0; totA=0; timesA=()
zeroQ=0; zeroH=0; low=0
declare -a events

for r in $(seq 1 "$ROUNDS"); do
    # 各自实例"最近 500 行里，注册后又没掉线"的净连接数 —— 用最近事件近似
    nq=$(tail -200 /root/lanremote-relay/tunnel.log 2>/dev/null \
         | grep -c 'Registered tunnel connection')
    nh=$(tail -200 /root/lanremote-relay/tunnel-h2.log 2>/dev/null \
         | grep -c 'Registered tunnel connection')

    # 每 3 轮（30 秒）打一次公网
    if [ $((r % 3)) -eq 0 ]; then
        out=$(curl -s -o /dev/null -w '%{http_code} %{time_total}' --max-time 25 \
              -H "User-Agent: $UA" "$BASE/v1/health")
        c=$(echo "$out" | awk '{print $1}')
        ms=$(echo "$out" | awk '{printf "%d", $2*1000}')
        totA=$((totA + 1))
        if [ "$c" = "200" ]; then
            okA=$((okA + 1)); timesA+=("$ms")
        else
            events+=("[$(date '+%H:%M:%S')] HTTP $c")
            printf "  [%s] ✗ HTTP %s   (Q=%s H=%s)\n" "$(date '+%H:%M:%S')" "$c" "$nq" "$nh" | tee -a "$OUT"
        fi
    fi
    sleep 10
done

echo | tee -a "$OUT"
echo "════════ 结果 ════════" | tee -a "$OUT"
pct=$((okA * 100 / (totA > 0 ? totA : 1)))
echo "  公网可用率: $okA/$totA = ${pct}%" | tee -a "$OUT"
if [ "${#timesA[@]}" -gt 0 ]; then
    printf '%s\n' "${timesA[@]}" | sort -n | awk '{
        a[NR]=$1
    } END {
        printf "  耗时: 最小 %d  中位 %d  P90 %d  P99 %d  最大 %d ms\n",
            a[1], a[int(NR/2)+1], a[int(NR*0.9)+1], a[int(NR*0.99)+1], a[NR]
    }' | tee -a "$OUT"
fi

echo | tee -a "$OUT"
echo "  QUIC 实例事件:" | tee -a "$OUT"
for kw in 'Registered tunnel connection' 'Lost connection' 'Failed to dial' 'no recent network activity'; do
    c=$(tail -n +$((MARK_Q+1)) /root/lanremote-relay/tunnel.log 2>/dev/null | grep -c "$kw" || echo 0)
    printf "    %-34s %s 次\n" "$kw" "$c" | tee -a "$OUT"
done
echo "  HTTP/2 实例事件:" | tee -a "$OUT"
for kw in 'Registered tunnel connection' 'Lost connection' 'Failed to dial' 'i/o timeout'; do
    c=$(tail -n +$((MARK_H+1)) /root/lanremote-relay/tunnel-h2.log 2>/dev/null | grep -c "$kw" || echo 0)
    printf "    %-34s %s 次\n" "$kw" "$c" | tee -a "$OUT"
done

echo | tee -a "$OUT"
echo "  HTTP 失败时刻:" | tee -a "$OUT"
if [ "${#events[@]}" -gt 0 ]; then
    printf '    %s\n' "${events[@]}" | tee -a "$OUT"
else
    echo "    一次都没有" | tee -a "$OUT"
fi
echo | tee -a "$OUT"
echo "  存档: $OUT" | tee -a "$OUT"