Skip to content

PasteFerry (隔空贴): clipboard and file sync across macOS, Windows and Linux. Direct-first, end-to-end encrypted. Optional official access: no server to host. Independent self-hosting remains available.

14.2 KB
Raw Blame
#!/usr/bin/env bash
# NAT 打洞实验室:用 Linux 网络命名空间与 iptables 搭出真实的网络拓扑,
# 验证两台设备最终走直连还是中转。
#
#   sudo ./tools/natlab/run.sh <natlab 可执行文件>
#
# 需要 root、iproute2、iptables。macOS 上用 tools/natlab/docker.sh,它会在容器里跑本脚本。
#
# 拓扑(全部用内网地址扮演公网,natlab 以 ProbeAllowPrivate 运行):
#
#   [a] 192.168.10.2 ── [rA] 出口 10.255.0.11–14 ──┐
#                                                  [inet] 信令 10.200.0.2  TURN 10.200.0.3
#   [b] 192.168.20.2 ── [rB] 出口 10.254.0.21–22 ──┘        STUN 10.200.1.1–4
#
# 10.200.1.1 扮演自己的服务器(信令服务器下发的 STUN),10.200.1.2–4 扮演公共 STUN。
# rA 按目标地址选出口,模拟公司双线、运营商地址池;入站只放行有连接记录的包,
# 与真实路由器一样「只认打过招呼的地址」。
#
# 局域网场景(没有服务器,见 setup_lan)另搭一套拓扑:
#
#   同一网段  [a] 192.168.10.2 ── [sw] 网桥 ── [b] 192.168.10.3
#   两个网段  [a] 192.168.10.2 ── [r] 路由器,不做 NAT ── [b] 192.168.20.2

set -euo pipefail

BIN=$(cd "$(dirname "${1:?用法: run.sh <natlab 可执行文件>}")" && pwd)/$(basename "$1")
WORK=$(mktemp -d)
PROBE="10.200.1.2:3478,10.200.1.3:3478,10.200.1.4:3478"

ns() { ip netns exec "$@"; }

INFRA_PID=""
INFRA_ARGS=() # 各场景额外传给 infra 的参数,由 check 设置

# 按进程号结束 infra:用 pkill -f 按路径匹配会连本脚本自己一起杀掉(命令行里也有这个路径)
cleanup() {
    if [[ -n $INFRA_PID ]]; then
        kill "$INFRA_PID" 2>/dev/null || true
        wait "$INFRA_PID" 2>/dev/null || true
        INFRA_PID=""
    fi
    for n in inet rA a rB b sw r; do ip netns del "$n" 2>/dev/null || true; done
}
trap 'cleanup; rm -rf "$WORK"' EXIT

veth() { # <命名空间1> <网卡1> <命名空间2> <网卡2>
    ip link add "$2" netns "$1" type veth peer name "$4" netns "$3"
}

# setup <A 侧出口模式> <B 侧模式>
#   A:single 单出口 | dual 两个出口,去对端走出口 2、探测服务器有一台走出口 2
#      hidden 两个出口,但没有任何探测服务器走出口 2 | quad 四个出口
#   B:cone 家用路由器 | symmetric 端口随机分配 | dual 两个出口
setup() {
    cleanup
    for n in inet rA a rB b; do
        ip netns add "$n"
        ns "$n" ip link set lo up
    done
    veth inet i-a rA wan
    veth inet i-b rB wan
    veth rA lan a eth0
    veth rB lan b eth0

    ns inet ip addr add 10.255.0.1/24 dev i-a
    ns inet ip addr add 10.254.0.1/24 dev i-b
    for ip in 10.200.0.2 10.200.0.3 10.200.1.1 10.200.1.2 10.200.1.3 10.200.1.4; do
        ns inet ip addr add "$ip/32" dev lo
    done
    ns inet ip link set i-a up
    ns inet ip link set i-b up
    ns inet sysctl -qw net.ipv4.ip_forward=1

    for i in 1 2 3 4; do ns rA ip addr add "10.255.0.1$i/24" dev wan; done
    ns rA ip addr add 192.168.10.1/24 dev lan
    ns rA ip link set wan up
    ns rA ip link set lan up
    ns rA ip route add default via 10.255.0.1
    ns rA sysctl -qw net.ipv4.ip_forward=1
    ns a ip addr add 192.168.10.2/24 dev eth0
    ns a ip link set eth0 up
    ns a ip route add default via 192.168.10.1

    for i in 1 2; do ns rB ip addr add "10.254.0.2$i/24" dev wan; done
    ns rB ip addr add 192.168.20.1/24 dev lan
    ns rB ip link set wan up
    ns rB ip link set lan up
    ns rB ip route add default via 10.254.0.1
    ns rB sysctl -qw net.ipv4.ip_forward=1
    ns b ip addr add 192.168.20.2/24 dev eth0
    ns b ip link set eth0 up
    ns b ip route add default via 192.168.20.1

    # 路由器的防火墙:丢弃外网主动发来的新连接,与真实的家用、公司路由器一致。
    # 不能省:被丢弃的包不会留下连接记录。若放行到路由器本机,Linux 会为它建一条
    # 「未应答」的连接记录,随后本端往外发时源端口与之冲突,SNAT 只好换一个端口,
    # 事先告诉对端的地址就失效了——真实路由器不会这样。
    for r in rA rB; do
        ns "$r" iptables -A INPUT -i wan -m conntrack --ctstate NEW -j DROP
        ns "$r" iptables -A FORWARD -i wan -m conntrack --ctstate NEW -j DROP
    done

    local snat="iptables -t nat -A POSTROUTING -o wan"
    case $1 in
    single) ns rA $snat -j SNAT --to-source 10.255.0.11 ;;
    dual)
        ns rA $snat -d 10.254.0.0/24 -j SNAT --to-source 10.255.0.12
        ns rA $snat -d 10.200.1.2 -j SNAT --to-source 10.255.0.12
        ns rA $snat -j SNAT --to-source 10.255.0.11
        ;;
    hidden)
        ns rA $snat -d 10.254.0.0/24 -j SNAT --to-source 10.255.0.12
        ns rA $snat -j SNAT --to-source 10.255.0.11
        ;;
    quad)
        ns rA $snat -d 10.254.0.0/24 -j SNAT --to-source 10.255.0.14
        for i in 2 3 4; do ns rA $snat -d "10.200.1.$i" -j SNAT --to-source "10.255.0.1$i"; done
        ns rA $snat -j SNAT --to-source 10.255.0.11
        ;;
    esac
    case $2 in
    cone) ns rB $snat -j SNAT --to-source 10.254.0.21 ;;
    symmetric) ns rB $snat -j SNAT --to-source 10.254.0.21 --random-fully ;;
    dual)
        ns rB $snat -d 10.255.0.0/24 -j SNAT --to-source 10.254.0.22
        ns rB $snat -d 10.200.1.3 -j SNAT --to-source 10.254.0.22
        ns rB $snat -j SNAT --to-source 10.254.0.21
        ;;
    esac

    ns inet "$BIN" infra -http 10.200.0.2:9000 -turn 10.200.0.3 \
        -stun 10.200.1.1,10.200.1.2,10.200.1.3,10.200.1.4 "${INFRA_ARGS[@]}" >"$WORK/infra.log" 2>&1 &
    INFRA_PID=$!
    for _ in $(seq 50); do
        ns a bash -c 'exec 3<>/dev/tcp/10.200.0.2/9000' 2>/dev/null && return
        sleep 0.1
    done
    echo "infra 没有起来:" && cat "$WORK/infra.log" && exit 1
}

PASS=0
FAIL=0

# check <场景名> <A 侧> <B 侧> <direct|relay> [legacy|history|late|leaky-a|leaky-b]
check() {
    # NATLAB_ONLY 只跑名字里含这段文字的场景,排查问题时用
    [[ -n ${NATLAB_ONLY:-} && $1 != *"$NATLAB_ONLY"* ]] && return
    local name=$1 expect=$4 flag=${5:-} aflags=() bflags=() state="$WORK/a-state.json"
    rm -f "$state" # 每个场景从空白开始:上一个场景的出口历史会让这一个「意外」直连
    INFRA_ARGS=()
    # 现实里信令经服务器绕一道,比设备之间的包慢几十毫秒:对端拿到本端地址就开始发包,
    # 本端要多等一个来回才拿到对端的地址。leaky 场景靠这个时间差复现问题
    [[ $flag == leaky-* ]] && INFRA_ARGS=(-delay 80ms)
    setup "$2" "$3"
    case $flag in
    legacy) aflags=(-legacy) ;;
    history)
        # 预置一份历史:这个网络里见过不改端口的出口 2
        printf '{"networks":{"192.168.10.2":[{"ip":"10.255.0.12","port_preserved":true,"last_seen":"%s"}]}}' \
            "$(date -u +%Y-%m-%dT%H:%M:%SZ)" >"$state"
        ;;
    late)
        # 开头 8 秒 rB 不放行发往 a 那边的 UDP,第一次握手只能落到中转;
        # 之后放开,a 应当用 ICE 重启换成直连。对应现实里「对端刚启动,
        # 出口地址晚到」「网络刚恢复」等一次没打通、之后其实打得通的情形
        local block=(FORWARD -i lan -d 10.255.0.0/24 -p udp -j DROP)
        ns rB iptables -I "${block[@]}"
        (sleep 8 && ns rB iptables -D "${block[@]}") &
        aflags=(-settle 40s)
        bflags=(-settle 40s)
        ;;
    leaky-a | leaky-b)
        # 路由器不丢弃外网主动发来的包:包落到路由器本机,留下一条连接记录。对端的包若先到,
        # 本端随后往同一地址发包时源端口被占,SNAT 只好换一个端口,事先告诉对端的地址就失效了。
        # 实测公司路由器就是这样:对端先发包的那几次都没打通。
        # leaky-a:a(发起方)的路由器这样,初次握手由 a 先发包,应当直接打通;
        # leaky-b:b(应答方)的路由器这样,初次握手打不通,第一次重试换 b 先发包后打通
        local r=rA
        [[ $flag == leaky-b ]] && r=rB
        ns "$r" iptables -D INPUT -i wan -m conntrack --ctstate NEW -j DROP
        aflags=(-settle 40s)
        bflags=(-settle 40s)
        ;;
    esac
    local common=(-infra http://10.200.0.2:9000 -stun 10.200.1.1:3478 -timeout 30s)
    ns b "$BIN" peer -id b -peer a "${common[@]}" -probe "$PROBE" "${bflags[@]}" >"$WORK/b.json" 2>"$WORK/b.log" &
    local bpid=$!
    ns a "$BIN" peer -id a -peer b "${common[@]}" -probe "$PROBE" -state "$state" "${aflags[@]}" \
        >"$WORK/a.json" 2>"$WORK/a.log" || true
    wait "$bpid" || true

    local a b
    a=$(sed -n 's/.*"state":"\([a-z]*\)".*/\1/p' "$WORK/a.json")
    b=$(sed -n 's/.*"state":"\([a-z]*\)".*/\1/p' "$WORK/b.json")
    local ok=0
    [[ $a == "$expect" && $b == "$expect" ]] && ok=1
    # 「晚到」场景必须真的先落到中转、再靠 ICE 重启换成直连,否则没测到要测的东西
    if [[ $flag == late ]] && ! grep -q '尝试换成直连' "$WORK/a.log"; then
        ok=0
        echo "    a 没有经过「中转 → 尝试换成直连」"
    fi
    if [[ $ok == 1 ]]; then
        PASS=$((PASS + 1))
        printf '  ✓ %-34s %s\n' "$name" "$a"
    else
        FAIL=$((FAIL + 1))
        printf '  ✗ %-34s 期望 %s,实际 a=%s b=%s\n' "$name" "$expect" "${a:-无结果}" "${b:-无结果}"
        echo "    a: $(cat "$WORK/a.json")"
        echo "    b: $(cat "$WORK/b.json")"
        grep -h -E '出口探测完成|P2P|换成直连|ICE 重启|失败|error' "$WORK/a.log" "$WORK/b.log" | head -8 | sed 's/^/    /'
    fi
    # NATLAB_KEEP 指定目录时把每个场景的完整日志留下来
    if [[ -n ${NATLAB_KEEP:-} ]]; then
        local dir="$NATLAB_KEEP/${name// /_}"
        mkdir -p "$dir" && cp "$WORK"/*.log "$dir/" 2>/dev/null || true
    fi
}

# setup_lan <same|routed|isolated>:没有服务器的局域网
#   same      a、b 接在同一台交换机(网桥)上,同一网段,mDNS 互通
#   routed    a、b 在两个网段,路由器转发、不做 NAT;mDNS 过不去
#   isolated  同 routed,但路由器不让两个网段互通(访客网络与办公网络常这样隔开)
setup_lan() {
    cleanup
    for n in a b; do
        ip netns add "$n"
        ns "$n" ip link set lo up
    done
    case $1 in
    same)
        ip netns add sw
        ns sw ip link add br0 type bridge
        ns sw ip link set br0 up
        veth sw p-a a eth0
        veth sw p-b b eth0
        for p in p-a p-b; do
            ns sw ip link set "$p" master br0
            ns sw ip link set "$p" up
        done
        ns a ip addr add 192.168.10.2/24 dev eth0
        ns b ip addr add 192.168.10.3/24 dev eth0
        for n in a b; do
            ns "$n" ip link set eth0 up
            ns "$n" ip route replace 224.0.0.0/4 dev eth0 # 没有默认路由,组播要有个去处
        done
        ;;
    routed | isolated)
        ip netns add r
        veth r la a eth0
        veth r lb b eth0
        ns r ip addr add 192.168.10.1/24 dev la
        ns r ip addr add 192.168.20.1/24 dev lb
        ns r ip link set la up
        ns r ip link set lb up
        ns r sysctl -qw net.ipv4.ip_forward=1
        ns a ip addr add 192.168.10.2/24 dev eth0
        ns b ip addr add 192.168.20.2/24 dev eth0
        ns a ip link set eth0 up
        ns b ip link set eth0 up
        ns a ip route add default via 192.168.10.1
        ns b ip route add default via 192.168.20.1
        if [[ $1 == isolated ]]; then
            ns r iptables -A FORWARD -j DROP
        fi
        ;;
    esac
}

# check_lan <场景名> <same|routed|isolated> <direct|offline> "<a 的参数>" "<b 的参数>" [hostfw]
#   hostfw:两台设备自己都拦截入站的新连接,模拟没有放行规则的 Windows 防火墙
check_lan() {
    [[ -n ${NATLAB_ONLY:-} && $1 != *"$NATLAB_ONLY"* ]] && return
    local name=$1 expect=$3
    setup_lan "$2"
    if [[ ${6:-} == hostfw ]]; then
        for n in a b; do ns "$n" iptables -A INPUT -i eth0 -p udp -m conntrack --ctstate NEW -j DROP; done
    fi
    # shellcheck disable=SC2086 # 参数故意按空格拆开
    ns b "$BIN" lanpeer -seed b -peer-seed a $5 >"$WORK/b.json" 2>"$WORK/b.log" &
    local bpid=$!
    # shellcheck disable=SC2086
    ns a "$BIN" lanpeer -seed a -peer-seed b $4 >"$WORK/a.json" 2>"$WORK/a.log" || true
    wait "$bpid" || true

    local a b
    a=$(sed -n 's/.*"state":"\([a-z]*\)".*/\1/p' "$WORK/a.json")
    b=$(sed -n 's/.*"state":"\([a-z]*\)".*/\1/p' "$WORK/b.json")
    if [[ $a == "$expect" && $b == "$expect" ]]; then
        PASS=$((PASS + 1))
        printf '  ✓ %-34s %s\n' "$name" "$a"
    else
        FAIL=$((FAIL + 1))
        printf '  ✗ %-34s 期望 %s,实际 a=%s b=%s\n' "$name" "$expect" "${a:-无结果}" "${b:-无结果}"
        grep -h -E '局域网|P2P|失败|error' "$WORK/a.log" "$WORK/b.log" | head -8 | sed 's/^/    /'
    fi
    if [[ -n ${NATLAB_KEEP:-} ]]; then
        local dir="$NATLAB_KEEP/${name// /_}"
        mkdir -p "$dir" && cp "$WORK"/*.log "$dir/" 2>/dev/null || true
    fi
}

echo "NAT 打洞实验室"
check "单出口 ↔ 家用路由器" single cone direct
check "两个出口(旧版本,对照)" dual cone relay legacy
check "两个出口 ↔ 家用路由器" dual cone direct
check "四个出口 ↔ 家用路由器" quad cone direct
check "两个出口,第二条没探测到,无历史" hidden cone relay
check "两个出口,第二条没探测到,有历史" hidden cone direct history
check "两个出口 ↔ 两个出口" dual dual direct
check "两个出口 ↔ 端口随机分配的 NAT" dual symmetric relay
check "一开始打不通,先中转后换直连" dual cone direct late
check "公司路由器会为外来的包占住端口" dual cone direct leaky-a
check "家用路由器会为外来的包占住端口" dual cone direct leaky-b

echo
echo "局域网(没有服务器)"
A=192.168.10.2:8788
B=192.168.20.2:8788
check_lan "同一网段,靠 mDNS 发现" same direct "-mdns" "-mdns"
check_lan "两个网段,一边记得对方地址" routed direct "-addr $B" ""
check_lan "两个网段,两台都挡入站,都记得地址" routed direct "-addr $B" "-addr $A" hostfw
check_lan "两个网段,两台都挡入站,一边记得" routed offline "-addr $B" "" hostfw
check_lan "两个网段之间被隔离" isolated offline "-addr $B" "-addr $A"

echo
echo "通过 ${PASS},失败 $FAIL"
[[ $FAIL -eq 0 ]]