当前位置: 首页 > news >正文

Netty - 从Nginx 四层(TCP/UDP)流量中获取客户端真实/网络出口IP

文章目录

  • 一、背景与原理
    • 1.1 问题场景
      • 网络架构影响分析
        • 1.1 客户端与Nginx之间存在的NAT/VPN
        • 1.2 Nginx与RPC服务之间的NAT
    • 1.2 技术原理
  • 二、环境配置验证
    • 2.1 Nginx配置
    • 2.2 版本要求
  • 三、Netty服务端实现
    • 3.1 Pipeline配置(核心代码)
    • 3.2 协议处理器实现
    • 3.3 业务处理器调用示例
  • 四、关键实现细节
    • 4.1 解码顺序保障
    • 4.2 地址类型处理
    • 4.3 安全性增强
  • 五、异常处理方案
  • 六、 HAProxyMessageDecoder
  • 七、总结

在这里插入图片描述

一、背景与原理

1.1 问题场景

在TCP四层代理场景下,Nginx作为反向代理将客户端请求转发至Netty实现的RPC服务。由于经过代理转发,RPC服务默认只能获取到Nginx服务器的IP地址。需要通过Proxy Protocol协议传递客户端真实IP。


网络架构影响分析

1.1 客户端与Nginx之间存在的NAT/VPN
场景可获取IP类型技术原理
企业级NAT网关仅NAT出口公网IPNAT设备替换源IP地址
家用路由器NAT路由器WAN口IP私有地址转换为公网IP
全局VPN接入VPN服务器出口IP流量封装后隧道传输

结论:Proxy Protocol只能传递Nginx直接看到的IP(即NAT/VPN出口地址)

1.2 Nginx与RPC服务之间的NAT
场景影响程度解决方案
普通NAT转发无影响保持现有Proxy Protocol配置
复杂SDN网络需验证确保TCP连接透传代理协议头

1.2 技术原理

  • Proxy Protocol:由HAProxy提出的传输层协议扩展,在建立TCP连接时发送包含源地址信息的头部
  • Nginx配置proxy_protocol on指令启用协议支持
  • Netty解码:通过HAProxyMessageDecoder解析协议头

二、环境配置验证

2.1 Nginx配置

仅用于演示,仅展示核心配置。

# 定义后端服务器组
upstream rpc_backend {
    server 10.0.0.1:12345;
    server 10.0.0.2:12345;
}

# 配置TCP代理
stream {
    server {
        listen 12345;
        proxy_pass rpc_backend;
        proxy_protocol on;
    }
}

在Nginx的stream模块中,设置一个TCP代理服务器,监听12345端口,将所有到达该端口的连接通过proxy protocol转发到rpc_backend定的后端服务器组。这样后端服务器可以获取到原始客户端的IP地址等信息,前提是后端服务支持proxy protocol。

proxy_protocol on

  • 启用PROXY协议(版本1或2),在转发流量时,将客户端的原始信息(如源IP、端口)附加到数据包头部。
  • 作用:使后端服务器能获取客户端真实IP,而非Nginx代理的IP。
  • 要求:后端服务必须支持并配置解析PROXY协议。

2.2 版本要求

  • Nginx ≥ 1.9.0 我这里是1.27.1
  • Netty ≥ 4.1.x 我这里是 4.1.109

三、Netty服务端实现

3.1 Pipeline配置(核心代码)

public class RpcServerInitializer extends ChannelInitializer<SocketChannel> {
    @Override
    protected void initChannel(SocketChannel ch) {
        ChannelPipeline pipeline = ch.pipeline();
        
        // 添加PROXY协议解码器
        pipeline.addLast(new HAProxyMessageDecoder());
        
        // 自定义协议处理器
        pipeline.addLast(new ProxyProtocolHandler());
        pipeline.addLast(new RpcMessageDecoder());
        pipeline.addLast(new RpcMessageHandler());
    }
}

3.2 协议处理器实现

提取IP

public class ProxyProtocolHandler extends ChannelInboundHandlerAdapter {
    
    @Override
    public void channelRead(ChannelHandlerContext ctx, Object msg) {
        if (msg instanceof HAProxyMessage) {
            HAProxyMessage proxyMessage = (HAProxyMessage) msg;
            
            // 提取真实客户端地址
            String clientIp = proxyMessage.sourceAddress();
            int clientPort = proxyMessage.sourcePort();
            
            // 存储到Channel属性中
            ctx.channel().attr(CLIENT_IP_ATTRIBUTE).set(clientIp);
            
            // 释放资源并移除当前消息
            ReferenceCountUtil.release(msg);
            return;
        }
        
        // 非PROXY协议消息继续传递
        ctx.fireChannelRead(msg);
    }

    @Override
    public void exceptionCaught(ChannelHandlerContext ctx, Throwable cause) {
        log.error("Proxy protocol parsing failed", cause);
        ctx.close();
    }
}

3.3 业务处理器调用示例

使用IP

public class RpcMessageHandler extends SimpleChannelInboundHandler<RpcRequest> {
    
    @Override
    protected void channelRead0(ChannelHandlerContext ctx, RpcRequest request) {
        String clientIp = ctx.channel().attr(CLIENT_IP_ATTRIBUTE).get();
        
        log.info("Received request from {}: {}", clientIp, request);
        // 业务处理逻辑...
    }
}

四、关键实现细节

4.1 解码顺序保障

  • HAProxyMessageDecoder必须作为第一个入站处理器
  • 需要处理完PROXY协议头后立即移除解码器(自动完成)

我抓了个包,如下

在这里插入图片描述

右键 Proxyv1 追踪流
在这里插入图片描述


4.2 地址类型处理

// 支持IPv4/IPv6地址类型判断
if (proxyMessage.proxyProtocol().addressType() == HAProxyAddressType.IPV4) {
    // IPv4处理逻辑
} else if (proxyMessage.proxyProtocol().addressType() == HAProxyAddressType.IPV6) {
    // IPv6处理逻辑
}

4.3 安全性增强

// 限制允许的代理服务器IP(可选)
List<String> allowedProxies = Arrays.asList("10.0.0.0/8", "192.168.0.0/16");
if (!isAllowedProxy(proxyMessage.destinationAddress())) {
    ctx.close();
    return;
}

五、异常处理方案

异常场景处理方案
无效PROXY头记录日志并关闭连接
协议版本不匹配返回错误响应码
地址格式错误使用默认地址并告警

六、 HAProxyMessageDecoder

/*
 * Copyright 2014 The Netty Project
 *
 * The Netty Project licenses this file to you under the Apache License,
 * version 2.0 (the "License"); you may not use this file except in compliance
 * with the License. You may obtain a copy of the License at:
 *
 *   https://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
 * License for the specific language governing permissions and limitations
 * under the License.
 */
package io.netty.handler.codec.haproxy;

import io.netty.buffer.ByteBuf;
import io.netty.channel.ChannelHandlerContext;
import io.netty.handler.codec.ByteToMessageDecoder;
import io.netty.handler.codec.ProtocolDetectionResult;
import io.netty.util.CharsetUtil;

import java.util.List;

import static io.netty.handler.codec.haproxy.HAProxyConstants.*;

/**
 * Decodes an HAProxy proxy protocol header
 *
 * @see <a href="https://haproxy.1wt.eu/download/1.5/doc/proxy-protocol.txt">Proxy Protocol Specification</a>
 */
public class HAProxyMessageDecoder extends ByteToMessageDecoder {
    /**
     * Maximum possible length of a v1 proxy protocol header per spec
     */
    private static final int V1_MAX_LENGTH = 108;

    /**
     * Maximum possible length of a v2 proxy protocol header (fixed 16 bytes + max unsigned short)
     */
    private static final int V2_MAX_LENGTH = 16 + 65535;

    /**
     * Minimum possible length of a fully functioning v2 proxy protocol header (fixed 16 bytes + v2 address info space)
     */
    private static final int V2_MIN_LENGTH = 16 + 216;

    /**
     * Maximum possible length for v2 additional TLV data (max unsigned short - max v2 address info space)
     */
    private static final int V2_MAX_TLV = 65535 - 216;

    /**
     * Binary header prefix length
     */
    private static final int BINARY_PREFIX_LENGTH = BINARY_PREFIX.length;

    /**
     * {@link ProtocolDetectionResult} for {@link HAProxyProtocolVersion#V1}.
     */
    private static final ProtocolDetectionResult<HAProxyProtocolVersion> DETECTION_RESULT_V1 =
            ProtocolDetectionResult.detected(HAProxyProtocolVersion.V1);

    /**
     * {@link ProtocolDetectionResult} for {@link HAProxyProtocolVersion#V2}.
     */
    private static final ProtocolDetectionResult<HAProxyProtocolVersion> DETECTION_RESULT_V2 =
            ProtocolDetectionResult.detected(HAProxyProtocolVersion.V2);

    /**
     * Used to extract a header frame out of the {@link ByteBuf} and return it.
     */
    private HeaderExtractor headerExtractor;

    /**
     * {@code true} if we're discarding input because we're already over maxLength
     */
    private boolean discarding;

    /**
     * Number of discarded bytes
     */
    private int discardedBytes;

    /**
     * Whether or not to throw an exception as soon as we exceed maxLength.
     */
    private final boolean failFast;

    /**
     * {@code true} if we're finished decoding the proxy protocol header
     */
    private boolean finished;

    /**
     * Protocol specification version
     */
    private int version = -1;

    /**
     * The latest v2 spec (2014/05/18) allows for additional data to be sent in the proxy protocol header beyond the
     * address information block so now we need a configurable max header size
     */
    private final int v2MaxHeaderSize;

    /**
     * Creates a new decoder with no additional data (TLV) restrictions, and should throw an exception as soon as
     * we exceed maxLength.
     */
    public HAProxyMessageDecoder() {
        this(true);
    }

    /**
     * Creates a new decoder with no additional data (TLV) restrictions, whether or not to throw an exception as soon
     * as we exceed maxLength.
     *
     * @param failFast Whether or not to throw an exception as soon as we exceed maxLength
     */
    public HAProxyMessageDecoder(boolean failFast) {
        v2MaxHeaderSize = V2_MAX_LENGTH;
        this.failFast = failFast;
    }

    /**
     * Creates a new decoder with restricted additional data (TLV) size, and should throw an exception as soon as
     * we exceed maxLength.
     * <p>
     * <b>Note:</b> limiting TLV size only affects processing of v2, binary headers. Also, as allowed by the 1.5 spec
     * TLV data is currently ignored. For maximum performance it would be best to configure your upstream proxy host to
     * <b>NOT</b> send TLV data and instantiate with a max TLV size of {@code 0}.
     * </p>
     *
     * @param maxTlvSize maximum number of bytes allowed for additional data (Type-Length-Value vectors) in a v2 header
     */
    public HAProxyMessageDecoder(int maxTlvSize) {
        this(maxTlvSize, true);
    }

    /**
     * Creates a new decoder with restricted additional data (TLV) size, whether or not to throw an exception as soon
     * as we exceed maxLength.
     *
     * @param maxTlvSize maximum number of bytes allowed for additional data (Type-Length-Value vectors) in a v2 header
     * @param failFast Whether or not to throw an exception as soon as we exceed maxLength
     */
    public HAProxyMessageDecoder(int maxTlvSize, boolean failFast) {
        if (maxTlvSize < 1) {
            v2MaxHeaderSize = V2_MIN_LENGTH;
        } else if (maxTlvSize > V2_MAX_TLV) {
            v2MaxHeaderSize = V2_MAX_LENGTH;
        } else {
            int calcMax = maxTlvSize + V2_MIN_LENGTH;
            if (calcMax > V2_MAX_LENGTH) {  // lgtm[java/constant-comparison]
                v2MaxHeaderSize = V2_MAX_LENGTH;
            } else {
                v2MaxHeaderSize = calcMax;
            }
        }
        this.failFast = failFast;
    }

    /**
     * Returns the proxy protocol specification version in the buffer if the version is found.
     * Returns -1 if no version was found in the buffer.
     */
    private static int findVersion(final ByteBuf buffer) {
        final int n = buffer.readableBytes();
        // per spec, the version number is found in the 13th byte
        if (n < 13) {
            return -1;
        }

        int idx = buffer.readerIndex();
        // 主要修改这里
        if (match(TEXT_PREFIX, buffer, idx)) {
            return 1;
        }
        if (match(BINARY_PREFIX, buffer, idx)) {
            return buffer.getByte(idx + BINARY_PREFIX_LENGTH);
        }
        return -1;
//        return match(BINARY_PREFIX, buffer, idx) ? buffer.getByte(idx + BINARY_PREFIX_LENGTH) : 1;
    }

    /**
     * Returns the index in the buffer of the end of header if found.
     * Returns -1 if no end of header was found in the buffer.
     */
    private static int findEndOfHeader(final ByteBuf buffer) {
        final int n = buffer.readableBytes();

        // per spec, the 15th and 16th bytes contain the address length in bytes
        if (n < 16) {
            return -1;
        }

        int offset = buffer.readerIndex() + 14;

        // the total header length will be a fixed 16 byte sequence + the dynamic address information block
        int totalHeaderBytes = 16 + buffer.getUnsignedShort(offset);

        // ensure we actually have the full header available
        if (n >= totalHeaderBytes) {
            return totalHeaderBytes;
        } else {
            return -1;
        }
    }

    /**
     * Returns the index in the buffer of the end of line found.
     * Returns -1 if no end of line was found in the buffer.
     */
    private static int findEndOfLine(final ByteBuf buffer) {
        final int n = buffer.writerIndex();
        for (int i = buffer.readerIndex(); i < n; i++) {
            final byte b = buffer.getByte(i);
            if (b == '\r' && i < n - 1 && buffer.getByte(i + 1) == '\n') {
                return i;  // \r\n
            }
        }
        return -1;  // Not found.
    }

    @Override
    public boolean isSingleDecode() {
        // ByteToMessageDecoder uses this method to optionally break out of the decoding loop after each unit of work.
        // Since we only ever want to decode a single header we always return true to save a bit of work here.
        return true;
    }

    @Override
    public void channelRead(ChannelHandlerContext ctx, Object msg) throws Exception {
        super.channelRead(ctx, msg);
        if (finished) {
            ctx.pipeline().remove(this);
        }
    }

    @Override
    protected final void decode(ChannelHandlerContext ctx, ByteBuf in, List<Object> out) throws Exception {
        // determine the specification version
        if (version == -1) {
            if ((version = findVersion(in)) == -1) {
                // 头部
                finished = true;
                return;
            }
        }

        ByteBuf decoded;

        if (version == 1) {
            decoded = decodeLine(ctx, in);
        } else {
            decoded = decodeStruct(ctx, in);
        }

        if (decoded != null) {
            finished = true;
            try {
                if (version == 1) {
                    out.add(HAProxyMessage.decodeHeader(decoded.toString(CharsetUtil.US_ASCII)));
                } else {
                    out.add(HAProxyMessage.decodeHeader(decoded));
                }
            } catch (HAProxyProtocolException e) {
                fail(ctx, null, e);
            }
        }
    }

    /**
     * Create a frame out of the {@link ByteBuf} and return it.
     *
     * @param ctx     the {@link ChannelHandlerContext} which this {@link HAProxyMessageDecoder} belongs to
     * @param buffer  the {@link ByteBuf} from which to read data
     * @return frame  the {@link ByteBuf} which represent the frame or {@code null} if no frame could
     *                be created
     */
    private ByteBuf decodeStruct(ChannelHandlerContext ctx, ByteBuf buffer) throws Exception {
        if (headerExtractor == null) {
            headerExtractor = new StructHeaderExtractor(v2MaxHeaderSize);
        }
        return headerExtractor.extract(ctx, buffer);
    }

    /**
     * Create a frame out of the {@link ByteBuf} and return it.
     *
     * @param ctx     the {@link ChannelHandlerContext} which this {@link HAProxyMessageDecoder} belongs to
     * @param buffer  the {@link ByteBuf} from which to read data
     * @return frame  the {@link ByteBuf} which represent the frame or {@code null} if no frame could
     *                be created
     */
    private ByteBuf decodeLine(ChannelHandlerContext ctx, ByteBuf buffer) throws Exception {
        if (headerExtractor == null) {
            headerExtractor = new LineHeaderExtractor(V1_MAX_LENGTH);
        }
        return headerExtractor.extract(ctx, buffer);
    }

    private void failOverLimit(final ChannelHandlerContext ctx, int length) {
        failOverLimit(ctx, String.valueOf(length));
    }

    private void failOverLimit(final ChannelHandlerContext ctx, String length) {
        int maxLength = version == 1 ? V1_MAX_LENGTH : v2MaxHeaderSize;
        fail(ctx, "header length (" + length + ") exceeds the allowed maximum (" + maxLength + ')', null);
    }

    private void fail(final ChannelHandlerContext ctx, String errMsg, Exception e) {
        finished = true;
        ctx.close(); // drop connection immediately per spec
        HAProxyProtocolException ppex;
        if (errMsg != null && e != null) {
            ppex = new HAProxyProtocolException(errMsg, e);
        } else if (errMsg != null) {
            ppex = new HAProxyProtocolException(errMsg);
        } else if (e != null) {
            ppex = new HAProxyProtocolException(e);
        } else {
            ppex = new HAProxyProtocolException();
        }
        throw ppex;
    }

    /**
     * Returns the {@link ProtocolDetectionResult} for the given {@link ByteBuf}.
     */
    public static ProtocolDetectionResult<HAProxyProtocolVersion> detectProtocol(ByteBuf buffer) {
        if (buffer.readableBytes() < 12) {
            return ProtocolDetectionResult.needsMoreData();
        }

        int idx = buffer.readerIndex();

        if (match(BINARY_PREFIX, buffer, idx)) {
            return DETECTION_RESULT_V2;
        }
        if (match(TEXT_PREFIX, buffer, idx)) {
            return DETECTION_RESULT_V1;
        }
        return ProtocolDetectionResult.invalid();
    }

    private static boolean match(byte[] prefix, ByteBuf buffer, int idx) {
        for (int i = 0; i < prefix.length; i++) {
            final byte b = buffer.getByte(idx + i);
            if (b != prefix[i]) {
                return false;
            }
        }
        return true;
    }

    /**
     * HeaderExtractor create a header frame out of the {@link ByteBuf}.
     */
    private abstract class HeaderExtractor {
        /** Header max size */
        private final int maxHeaderSize;

        protected HeaderExtractor(int maxHeaderSize) {
            this.maxHeaderSize = maxHeaderSize;
        }

        /**
         * Create a frame out of the {@link ByteBuf} and return it.
         *
         * @param ctx     the {@link ChannelHandlerContext} which this {@link HAProxyMessageDecoder} belongs to
         * @param buffer  the {@link ByteBuf} from which to read data
         * @return frame  the {@link ByteBuf} which represent the frame or {@code null} if no frame could
         *                be created
         * @throws Exception if exceed maxLength
         */
        public ByteBuf extract(ChannelHandlerContext ctx, ByteBuf buffer) throws Exception {
            final int eoh = findEndOfHeader(buffer);
            if (!discarding) {
                if (eoh >= 0) {
                    final int length = eoh - buffer.readerIndex();
                    if (length > maxHeaderSize) {
                        buffer.readerIndex(eoh + delimiterLength(buffer, eoh));
                        failOverLimit(ctx, length);
                        return null;
                    }
                    ByteBuf frame = buffer.readSlice(length);
                    buffer.skipBytes(delimiterLength(buffer, eoh));
                    return frame;
                } else {
                    final int length = buffer.readableBytes();
                    if (length > maxHeaderSize) {
                        discardedBytes = length;
                        buffer.skipBytes(length);
                        discarding = true;
                        if (failFast) {
                            failOverLimit(ctx, "over " + discardedBytes);
                        }
                    }
                    return null;
                }
            } else {
                if (eoh >= 0) {
                    final int length = discardedBytes + eoh - buffer.readerIndex();
                    buffer.readerIndex(eoh + delimiterLength(buffer, eoh));
                    discardedBytes = 0;
                    discarding = false;
                    if (!failFast) {
                        failOverLimit(ctx, "over " + length);
                    }
                } else {
                    discardedBytes += buffer.readableBytes();
                    buffer.skipBytes(buffer.readableBytes());
                }
                return null;
            }
        }

        /**
         * Find the end of the header from the given {@link ByteBuf},the end may be a CRLF, or the length given by the
         * header.
         *
         * @param buffer the buffer to be searched
         * @return {@code -1} if can not find the end, otherwise return the buffer index of end
         */
        protected abstract int findEndOfHeader(ByteBuf buffer);

        /**
         * Get the length of the header delimiter.
         *
         * @param buffer the buffer where delimiter is located
         * @param eoh index of delimiter
         * @return length of the delimiter
         */
        protected abstract int delimiterLength(ByteBuf buffer, int eoh);
    }

    private final class LineHeaderExtractor extends HeaderExtractor {

        LineHeaderExtractor(int maxHeaderSize) {
            super(maxHeaderSize);
        }

        @Override
        protected int findEndOfHeader(ByteBuf buffer) {
            return findEndOfLine(buffer);
        }

        @Override
        protected int delimiterLength(ByteBuf buffer, int eoh) {
            return buffer.getByte(eoh) == '\r' ? 2 : 1;
        }
    }

    private final class StructHeaderExtractor extends HeaderExtractor {

        StructHeaderExtractor(int maxHeaderSize) {
            super(maxHeaderSize);
        }

        @Override
        protected int findEndOfHeader(ByteBuf buffer) {
            return HAProxyMessageDecoder.findEndOfHeader(buffer);
        }

        @Override
        protected int delimiterLength(ByteBuf buffer, int eoh) {
            return 0;
        }
    }
}


七、总结

在存在NAT或VPN的网络架构中,通过Proxy Protocol获取客户端真实IP的能力受限于网络设备的位置。

若NAT/VPN位于客户端与Nginx之间(如企业VPN或家庭路由),Proxy Protocol仅能传递经过NAT转换或VPN隧道出口的IP(如公网IP或VPN分配地址),无法穿透获取终端设备的内网真实IP。若需突破此限制,可采取混合方案:客户端主动上报IP(需改造客户端代码)并配合网络设备改造(如VPN网关记录原始IP、专用隧道协议)。但需注意隐私合规风险,避免采集敏感信息。

对于常规业务场景,Proxy Protocol结合Nginx配置已能满足“获取客户端侧网络出口IP”的需求,但若涉及终端设备溯源,需结合应用层协议与网络基础设施深度协同实现。

在这里插入图片描述

http://www.dtcms.com/a/97162.html

相关文章:

  • Java实现pdf中动态插入图片
  • 如何在 Postman 中正确设置 Session 以维持用户状态?
  • 亚马逊云科技提供完全托管的DeepSeek-R1模型
  • SEO(搜索引擎优化)详解
  • 处理脚本中函数调用的异常
  • 基于深度强化学习的智能机器人路径规划技术研究
  • 第六届 蓝桥杯 嵌入式 省赛
  • Postman CORS 测试完全指南:轻松模拟跨域请求,排查 CORS 相关问题
  • 软考中级-软件设计师 23种设计模式(内含详细解析)
  • Gateway实战(二)、负载均衡
  • React 中shouldComponentUpdate生命周期方法的作用,如何利用它优化组件性能?
  • Python爬虫如何检测请求频率?
  • Docker-Volume数据卷详讲
  • 循环神经网络 - 给网络增加记忆能力
  • 优化webpack打包体积思路
  • WebSocket:实时双向通信技术详解与实战示例优化指南
  • Linux内核禁止_开启中断和处理器间中断
  • 鸿蒙前后端项目源码-点餐v3.0-原创!原创!原创!
  • js关于for of 与for in
  • webpack和vite之间的区别
  • 从 Word 到 HTML:使用 Aspose.Words 轻松实现 Word 文档的高保真转换
  • Linux:基础IO---缓冲区
  • 1688商品详情接口:深度解析与应用实践
  • 基于社交裂变的S2B2C电商模式创新研究——以“颜值PK+礼品卡+AI智能名片“融合生态为例
  • JDK 17 + Spring Boot 3 全栈升级实战指南--从语法革新到云原生,解锁企业级开发新范式
  • 学习openfoam的两个地方(案例和源码)
  • MATLAB中构建模式表达式
  • Python 在Word中查找并替换文本
  • git在实践使用中的操作流程
  • 力扣:回溯算法