详解C/C++如何发送与接收Kafka消息

一、背景

在实际工程中，难免会遇到不通系统之间通信，如何进行系统之间通信呢？(作为一个“全栈工程师”，必须要解决它！)。

系统之间通信方式很多如：系统之间调用（http/rpc等），异步间接调用如发送消息、公共存储等。目前，本人从事的项目中遇到web业务工程（Java）依赖与算法工程(C++) 处理的视频/图片分类与标记结果。两个系统之前数据通信采用了kafka消息方式。

算法工程为C/C++工程，本文将介绍如何在C/C++中如何发送与接收Kakfa消息（包含：Kafka的SASL认证方式），并提供了详细的源码和讲解。（至于Java中如何发送与接收Kakfa消息如有需要，可留言或私聊！）

二、环境依赖安装

# 下载librdkafka
git clone https://github.com/edenhill/librdkafka.git
 
# 编译
cd librdkafka
./configure --prefix=/usr/local
 
# 安装
sudo make install
 
# 验证：查看/usr/local/lib目录下是否有librdkafka文件
ls /usr/local/lib | grep kafka

三、编写kakfa生产者消费者

3.1 生产者

#include <rdkafka.h>     // 包含C API头文件
#include <iostream>
#include <cstring>
#include <cerrno>
  
int main() {
    const char *brokers = "xx.xx.xx.xx:7091"; // Kafka broker地址
    const char *topic_name = "kafka_msg_topic_test";
    const char *payload = "Hello, Kafka from librdkafka!";
    size_t len = strlen(payload);
  
    // 创建配置对象
    rd_kafka_conf_t *conf = rd_kafka_conf_new();
    if (!conf) {
        std::cerr << "Failed to create configuration object: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        return 1;
    }
  
    // 设置broker地址
    if (rd_kafka_conf_set(conf, "bootstrap.servers", brokers, NULL, 0) != RD_KAFKA_CONF_OK) {
        std::cerr << "Failed to set bootstrap.servers: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
  
    // 创建生产者实例
    rd_kafka_t *rk = rd_kafka_new(RD_KAFKA_PRODUCER, conf, NULL, 0);
    if (!rk) {
        std::cerr << "Failed to create producer: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
  
    // 创建topic句柄（可选，但推荐）
    rd_kafka_topic_t *rkt = rd_kafka_topic_new(rk, topic_name, NULL);
    if (!rkt) {
        std::cerr << "Failed to create topic handle: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_destroy(rk);
//        rd_kafka_conf_destroy(conf);
        return 1;
    }
  
    // 发送消息
    int32_t partition = RD_KAFKA_PARTITION_UA; // 自动选择分区
    int err = rd_kafka_produce(rkt, partition, RD_KAFKA_MSG_F_COPY, const_cast<char *>(payload), len, NULL, 0, NULL);
    if (err != RD_KAFKA_RESP_ERR_NO_ERROR) {
        std::cerr << "Failed to produce to topic " << topic_name << ": " << err << std::endl;
    } else {
        std::cout << "Produced " << len << " bytes to topic " << topic_name << std::endl;
    }
  
    // 等待所有消息发送完成（可选，但推荐）
    // 在实际生产代码中，您可能需要更复杂的逻辑来处理消息的发送和确认
    int msgs_sent = 0;
    while (rd_kafka_outq_len(rk) > 0) {
        rd_kafka_poll(rk, 100); // 轮询Kafka队列，直到所有消息都发送出去
        msgs_sent += rd_kafka_outq_len(rk);
    }
  
    // 销毁topic句柄
    rd_kafka_topic_destroy(rkt);
  
    // 销毁生产者实例
    rd_kafka_destroy(rk);
  
    // 销毁配置对象
//    rd_kafka_conf_destroy(conf);
  
    return 0;
}

3.2 消费者

#include <rdkafka.h>
#include <iostream>
#include <cerrno>
#include <cstring>
#include <cstdlib>
  
void error_cb(rd_kafka_t *rk, int err, const char *reason, void *opaque) {
    // 错误处理回调
    std::cerr << "Kafka error: " << err << ": " << reason << std::endl;
}
  
int main() {
    std::cerr << "start " << std::endl;
    const char *brokers = "xx.xx.xx.xx:7091"; // Kafka broker地址
    const char *group_id = "kafka_msg_topic_test"; // 消费者组ID
    const char *topic_name = "kafka_msg_topic_test"; // Kafka topic名称
    
    // 创建配置对象
    rd_kafka_conf_t *conf = rd_kafka_conf_new();
    if (!conf) {
        std::cerr << "Failed to create configuration object: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        return 1;
    }
    
    // 设置broker地址
    if (rd_kafka_conf_set(conf, "bootstrap.servers", brokers, NULL, 0) != RD_KAFKA_CONF_OK) {
        std::cerr << "Failed to set bootstrap.servers: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
    
    // 设置消费者组ID
    if (rd_kafka_conf_set(conf, "group.id", group_id, NULL, 0) != RD_KAFKA_CONF_OK) {
        std::cerr << "Failed to set group.id: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
    
    // 设置错误处理回调（可选）
    rd_kafka_conf_set_error_cb(conf, error_cb);
    
    // 创建消费者实例
    rd_kafka_t *rk = rd_kafka_new(RD_KAFKA_CONSUMER, conf, NULL, 0);
    if (!rk) {
        std::cerr << "Failed to create consumer: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        return 1;
    }
    
    
    // 创建一个topic分区列表
    rd_kafka_topic_partition_list_t *topics = rd_kafka_topic_partition_list_new(1);
    if (!topics) {
        std::cerr << "Failed to create topic partition list: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_destroy(rk);
        return 1;
    }
    
    // 添加topic到分区列表
    if (!rd_kafka_topic_partition_list_add(topics, topic_name, RD_KAFKA_PARTITION_UA)) {
        std::cerr << "Failed to add topic to partition list: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_topic_partition_list_destroy(topics);
        rd_kafka_destroy(rk);
        return 1;
    }
    // 订阅topic
    rd_kafka_resp_err_t err = rd_kafka_subscribe(rk, topics);
    if (err != RD_KAFKA_RESP_ERR_NO_ERROR) {
        std::cerr << "Failed to subscribe to topic: " << rd_kafka_err2str(err) << std::endl;
        rd_kafka_topic_partition_list_destroy(topics);
        rd_kafka_destroy(rk);
        return 1;
    }
    
    // 销毁分区列表（订阅后不再需要）
    rd_kafka_topic_partition_list_destroy(topics);
    
    
    // 轮询消息
    while (true) {
        rd_kafka_message_t *rkmessage;
        rkmessage = rd_kafka_consumer_poll(rk, 1000); // 等待1秒以获取消息
        
        if (rkmessage == NULL) {
            // 没有消息或者超时
            continue;
        }
        
        if (rkmessage->err) {
            // 处理错误
            if (rkmessage->err == RD_KAFKA_RESP_ERR__PARTITION_EOF) {
                // 消息流的末尾
                std::cout << "End of partition event" << std::endl;
            } else {
                // 打印错误并退出
                std::cerr << "Kafka consumer error: " << rd_kafka_message_errstr(rkmessage) << std::endl;
                break;
            }
        } else {
            // 处理消息
            std::cout << "Received message at offset " << rkmessage->offset
            << " from partition " << rkmessage->partition
            << " with key \"" << rkmessage->key << "\" and payload size "<< rkmessage->len
            << " value :" <<(char *)rkmessage->payload
            << std::endl;
            
            // 如果需要，可以在这里处理消息内容
            // 例如，使用rkmessage->payload()获取消息内容
            
            // 释放消息
            rd_kafka_message_destroy(rkmessage);
        }
    }
    
    // 清理
    rd_kafka_destroy(rk);
    
    return 0;
}

3.3 编译运行

3.3.1 编译生产者消费者

g++ -o send_kafka SendKakfaMessage.cpp -I/usr/local/include/librdkafka -lrdkafka++ -lrdkafka -lpthread

g++ -o receive_kafka ReceiveKafkaMessage.cpp -I/usr/local/include/librdkafka -lrdkafka++ -lrdkafka -lpthread

3.3.2 运行验证

执行时，若出现错误： error while loading shared libraries: librdkafka++.so.1: cannot open shared object file: No such file or directory

则需要执行下面环境变量配置：

export LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH

生产者：发送消息

消费者：接收消息

3.4 SASL认证kakfa

下面是，支持sasl认证的kakka生产者完整代码

#include <rdkafka.h>
#include <iostream>
#include <cstring>
#include <cerrno>
 
int main(int argc, char *argv[]) {
    const char *brokers = "xx.xx.xx.xx:8092"; // Kafka broker地址
    const char *username = "xxx";
    const char *password = "xxx";
    const char *topic_name = "kafka_msg_test_sasl";
    const char *payload = "Hello, Kafka from librdkafka! sasl";
    size_t len = strlen(payload);
     // 初始化配置
    rd_kafka_conf_t *conf = rd_kafka_conf_new();
    if (!conf)
    {
        std::cerr << "Failed to create configuration object: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        return 1;
    }
    char errstr[512]; // 声明一个足够大的字符数组来存储错误信息
 
    // 设置SASL相关的配置
    if (rd_kafka_conf_set(conf, "security.protocol", "SASL_PLAINTEXT", errstr, sizeof(errstr)) != RD_KAFKA_CONF_OK)
    {
        std::cerr << "Failed to set security.protocol: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
    if (rd_kafka_conf_set(conf, "sasl.mechanisms", "PLAIN", errstr, sizeof(errstr)) != RD_KAFKA_CONF_OK)
    {
        std::cerr << "Failed to set sasl.mechanisms: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
    if (rd_kafka_conf_set(conf, "sasl.username", username, errstr, sizeof(errstr)) != RD_KAFKA_CONF_OK)
    {
        std::cerr << "Failed to set sasl.username: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
    if (rd_kafka_conf_set(conf, "sasl.password", password, errstr, sizeof(errstr)) != RD_KAFKA_CONF_OK)
    {
        std::cerr << "Failed to set sasl.password: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
 
    if (rd_kafka_conf_set(conf, "bootstrap.servers", brokers, errstr, sizeof(errstr)) != RD_KAFKA_CONF_OK)
    {
        std::cerr << "Failed to set bootstrap.servers: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_conf_destroy(conf);
        return 1;
    }
    // 检查配置是否设置成功
    if (rd_kafka_conf_set(conf, "security.protocol", "SASL_PLAINTEXT", errstr, sizeof(errstr)) != RD_KAFKA_CONF_OK) {
        std::cerr << "Failed to set configuration: " << errstr << std::endl;
        return 1;
    }
 
    // 创建producer实例
    rd_kafka_t *rk = rd_kafka_new(RD_KAFKA_PRODUCER, conf, errstr, sizeof(errstr));
    if (!rk) {
        std::cerr << "Failed to create new producer: " << errstr << std::endl;
        return 1;
    }
 
    // 创建topic句柄（可选，但推荐）
    rd_kafka_topic_t *rkt = rd_kafka_topic_new(rk, topic_name, NULL);
    if (!rkt)
    {
        std::cerr << "Failed to create topic handle: " << rd_kafka_err2str(rd_kafka_last_error()) << std::endl;
        rd_kafka_destroy(rk);
        //        rd_kafka_conf_destroy(conf);
        return 1;
    }
 
    // 发送消息
    int32_t partition = RD_KAFKA_PARTITION_UA; // 自动选择分区
    int err = rd_kafka_produce(rkt, partition, RD_KAFKA_MSG_F_COPY, const_cast<char *>(payload), len, NULL, 0, NULL);
    if (err != RD_KAFKA_RESP_ERR_NO_ERROR)
    {
        std::cerr << "Failed to produce to topic " << topic_name << ": " << err << std::endl;
    }
    else
    {
        std::cout << "Produced " << len << " bytes to topic " << topic_name << std::endl;
    }
 
    // 等待所有消息发送完成（可选，但推荐）
    // 在实际生产代码中，您可能需要更复杂的逻辑来处理消息的发送和确认
    int msgs_sent = 0;
    while (rd_kafka_outq_len(rk) > 0)
    {
        rd_kafka_poll(rk, 100); // 轮询Kafka队列，直到所有消息都发送出去
        msgs_sent += rd_kafka_outq_len(rk);
    }
 
    // 销毁topic句柄
    rd_kafka_topic_destroy(rkt);
 
    // 清理资源
    rd_kafka_destroy(rk);
    return 0;
}

在kafka map 管理界面中查看发送效果如下：

3.5 结束语