Kafka和Spring集成实践

目录

---

1. 安装Zookeeper
2. 安装Kafka
3. 创建一个Spring项目
4. 使用Producer API发送消息到Kafka
5. 使用Kafka High Level API接收消息
6. 使用spring-integration-kafka发送消息

7. 使用spring-integration-kafka接收消息

   本文以单机的环境演示如何将Kafka和Spring集成。
   单机的环境最容易搭建， 并且只需在自己的PC上运行即可， 不需要很多的硬件环境，便于学习。 况且，本文的目的不是搭建ZooKeeper的集群环境， 而是重点介绍Kafka和Spring的应用。
   具体的软件环境如下：

OS: CentOS 6.4
Zookepper： zookeeper-3.4.6
Kafka： kafka_2.9.1-0.8.2-beta
Java： JDK 1.7.0_45-b18
Spring:4.0.6
本例子在我的这个环境中运行正常， 全部代码可以到 github 下载。

本文所有的操作系统用户都是root。 实际产品中可能安全标准需要特定的用户如zookeeper, kafka等。

安装Zookeeper

---

首先下载解压zookeeper,选择合适的镜像站点以加快下载速度。
我们可以将zookeeper加到系统服务中， 增加一个/etc/init.d/zookeeper文件。

cd /opt
wget  http://apache.fayea.com/apache-mirror/zookeeper/zookeeper-3.4.6/zookeeper-3.4.6.tar.gz
tar zxvf zookeeper-3.4.6.tar.gz
vi /etc/init.d/zookeeper

将https://raw.githubusercontent.com/apache/zookeeper/trunk/src/packages/rpm/init.d/zookeeper文件的内容拷贝到这个文件，修改其中的运行zookeeper的用户以及zookeeper的文件夹位置。

......
start() {
  echo -n $"Starting $desc (zookeeper): "
  daemon --user root /opt/zookeeper-3.4.6/zkServer.sh start
  RETVAL=$?
  echo
  [ $RETVAL -eq 0 ] && touch /var/lock/subsys/zookeeper
  return $RETVAL
}
stop() {
  echo -n $"Stopping $desc (zookeeper): "
  daemon --user root /opt/zookeeper-3.4.6/zkServer.sh stop
  RETVAL=$?
  sleep 5
  echo
  [ $RETVAL -eq 0 ] && rm -f /var/lock/subsys/zookeeper $PIDFILE
}
......

chmod 755 /etc/init.d/zookeeper
service zookeeper start

如果你不想加到服务，也可以直接运行zookeeper。

/opt/zookeeper-3.4.6/zkServer.sh start

安装Kafka

---

从合适的镜像站点下载最新的kafka并解压。

wget http://apache.01link.hk/kafka/0.8.2-beta/kafka_2.9.1-0.8.2-beta.tgz
tar zxvf kafka_2.9.1-0.8.2-beta.tgz
cd kafka_2.9.1-0.8.2-beta

启动Kafka:

bin/kafka-server-start.sh config/server.properties

创建一个test的topic：

bin/kafka-topics.sh --create --zookeeper localhost:2181 --replication-factor 1 --partitions 1 --topic test

可以利用kafka的命令启动一个生产者和消费者试验一下：

bin/kafka-console-producer.sh --broker-list localhost:9092 --topic test
This is a message
This is another message

bin/kafka-console-consumer.sh --zookeeper localhost:2181 --topic test --from-beginning
This is a message
This is another message

更多的介绍可以查看我翻译整理的 Kafka快速入门

创建一个Spring项目

---

以上的准备环境完成，让我们开始创建一个项目。
以前我写过一篇简单介绍: Spring 集成 Kafka.
spring-integration-kafka这个官方框架我就不介绍了。 我们主要使用它做集成。

首先我们先看一下使用Kafka自己的Producer/Consumer API发送/接收消息的例子。

使用Producer API发送消息到Kafka

OK，现在我们先看一个使用Kafka 自己的producer API发送消息的例子：

public class NativeProducer {
    public static void main(String[] args) {
        String topic= "test";
        long events = 100;
        Random rand = new Random();

        Properties props = new Properties();
        props.put("metadata.broker.list", "localhost:9092");
        props.put("serializer.class", "kafka.serializer.StringEncoder");
        props.put("request.required.acks", "1");

        ProducerConfig config = new ProducerConfig(props);

        Producer<String, String> producer = new Producer<String, String>(config);

        for (long nEvents = 0; nEvents < events; nEvents++) {                
               String msg = "NativeMessage-" + rand.nextInt() ; 
               KeyedMessage<String, String> data = new KeyedMessage<String, String>(topic, nEvents + "", msg);
               producer.send(data);
        }
        producer.close();
    }
}

这个例子中首先初始化Producer对象，指定相应的broker和serializer， 然后发送100个字符串消息给Kafka。

运行mvn package编译代码，执行查看结果：

java -cp target/lib/*:target/spring-kafka-demo-0.2.0-SNAPSHOT.jar com.colobu.spring_kafka_demo.NativeProducer

上面的消费者控制台窗口会打印出收到的消息:

......
NativeMessage--1645592376
NativeMessage-534168193
NativeMessage--1899432197
NativeMessage-1642480773
NativeMessage--911267171
NativeMessage-251458151
NativeMessage--55710397
NativeMessage-455515562
NativeMessage-1108982916
NativeMessage--1710296834
NativeMessage-2102648373
NativeMessage-499979365
NativeMessage--1200107003
NativeMessage-1184836299
NativeMessage--1161123005
NativeMessage-912582115
NativeMessage--1557863408
NativeMessage--1036456356
......

使用Kafka High Level API接收消息

用High level Consumer API接收消息

import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.Properties;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import kafka.consumer.ConsumerConfig;
import kafka.consumer.ConsumerIterator;
import kafka.consumer.KafkaStream;
import kafka.javaapi.consumer.ConsumerConnector;
public class NativeConsumer {
    private final ConsumerConnector consumer;
    private final String topic;
    private ExecutorService executor;
    public NativeConsumer(String a_zookeeper, String a_groupId, String a_topic) {
        consumer = kafka.consumer.Consumer.createJavaConsumerConnector(createConsumerConfig(a_zookeeper, a_groupId));
        this.topic = a_topic;
    }
    public void shutdown() {
        if (consumer != null)
            consumer.shutdown();
        if (executor != null)
            executor.shutdown();
    }
    public void run(int a_numThreads) {
        Map topicCountMap = new HashMap();
        topicCountMap.put(topic, new Integer(a_numThreads));
        Mapbyte[], byte[]>>> consumerMap = consumer.createMessageStreams(topicCountMap);
        Listbyte[], byte[]>> streams = consumerMap.get(topic);
        // now launch all the threads
        //
        executor = Executors.newFixedThreadPool(a_numThreads);
        // now create an object to consume the messages
        //
        int threadNumber = 0;
        for (final KafkaStream stream : streams) {
            executor.submit(new ConsumerTest(stream, threadNumber));
            threadNumber++;
        }
    }
    private static ConsumerConfig createConsumerConfig(String a_zookeeper, String a_groupId) {
        Properties props = new Properties();
        props.put("zookeeper.connect", a_zookeeper);
        props.put("group.id", a_groupId);
        props.put("zookeeper.session.timeout.ms", "400");
        props.put("zookeeper.sync.time.ms", "200");
        props.put("auto.commit.interval.ms", "1000");
        return new ConsumerConfig(props);
    }
    public static void main(String[] args) {
        String zooKeeper = "localhost:2181";
        String groupId = "mygroup";
        String topic = "test";
        int threads = 1;
        NativeConsumer example = new NativeConsumer(zooKeeper, groupId, topic);
        example.run(threads);
        try {
            Thread.sleep(10000);
        } catch (InterruptedException ie) {
        }
        //example.shutdown();
    }
}
class ConsumerTest implements Runnable {
    private KafkaStream m_stream;
    private int m_threadNumber;

    public ConsumerTest(KafkaStream a_stream, int a_threadNumber) {
        m_threadNumber = a_threadNumber;
        m_stream = a_stream;
    }

    public void run() {
        ConsumerIterator<byte[], byte[]> it = m_stream.iterator();
        while (it.hasNext())
            System.out.println("Thread " + m_threadNumber + ": " + new String(it.next().message()));
        System.out.println("Shutting down Thread: " + m_threadNumber);
    }
}

在生产者控制台输入几条消息，可以看到运行这个例子的控制台可以将这些消息打印出来。

教程的代码中还包括一个使用Simple Consumer API接收消息的例子。 因为spring-integration-kafka不支持这种API，这里也不列出对比代码了。

使用spring-integration-kafka发送消息

Outbound Channel Adapter用来发送消息到Kafka。 消息从Spring Integration Channel中读取。 你可以在Spring application context指定这个channel。
一旦配置好这个Channel，就可以利用这个Channel往Kafka发消息。 明显地，Spring Integration特定的消息发送给这个Adaptor，然后发送前在内部被转为Kafka消息。当前的版本要求你必须指定消息key和topic作为头部数据 (header)，消息作为有载荷(payload)。
例如

final MessageChannel channel = ctx.getBean("inputToKafka", MessageChannel.class);
channel.send(
        MessageBuilder.withPayload(payload) //设置有效载荷
                .setHeader("messageKey", "key") //指定key
                .setHeader("topic", "test").build()); /指定topic/

实际代码如下：

import java.util.Random;
import org.springframework.context.support.ClassPathXmlApplicationContext;
import org.springframework.integration.support.MessageBuilder;
import org.springframework.messaging.MessageChannel;
public class Producer {
    private static final String CONFIG = "/context.xml";
    private static Random rand = new Random();
    public static void main(String[] args) {
        final ClassPathXmlApplicationContext ctx = new ClassPathXmlApplicationContext(CONFIG, Producer.class);
        ctx.start();
        final MessageChannel channel = ctx.getBean("inputToKafka", MessageChannel.class);
        for (int i = 0; i < 100; i++) {
            channel.send(MessageBuilder.withPayload("Message-" + rand.nextInt()).setHeader("messageKey", String.valueOf(i)).setHeader("topic", "test").build());
        }
        try {
            Thread.sleep(100000);
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
        ctx.close();
    }
}

Spring 配置文件：

<beans xmlns="http://www.springframework.org/schema/beans"
       xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
       xmlns:int="http://www.springframework.org/schema/integration"
       xmlns:int-kafka="http://www.springframework.org/schema/integration/kafka"
       xmlns:task="http://www.springframework.org/schema/task"
       xsi:schemaLocation="http://www.springframework.org/schema/integration/kafka http://www.springframework.org/schema/integration/kafka/spring-integration-kafka.xsd
        http://www.springframework.org/schema/integration http://www.springframework.org/schema/integration/spring-integration.xsd
        http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd
        http://www.springframework.org/schema/task http://www.springframework.org/schema/task/spring-task.xsd">
    <int:channel id="inputToKafka">
        <int:queue/>
    int:channel>
    <int-kafka:outbound-channel-adapter id="kafkaOutboundChannelAdapter"
                                        kafka-producer-context-ref="kafkaProducerContext"
                                        auto-startup="false"
                                        channel="inputToKafka"
                                        order="3"
            >
        <int:poller fixed-delay="1000" time-unit="MILLISECONDS" receive-timeout="0" task-executor="taskExecutor"/>
    int-kafka:outbound-channel-adapter>
    <task:executor id="taskExecutor" pool-size="5" keep-alive="120" queue-capacity="500"/>
    <bean id="producerProperties"
        class="org.springframework.beans.factory.config.PropertiesFactoryBean">
        <property name="properties">
            <props>
                <prop key="topic.metadata.refresh.interval.ms">3600000prop>
                <prop key="message.send.max.retries">5prop>
                <prop key="serializer.class">kafka.serializer.StringEncoderprop>
                <prop key="request.required.acks">1prop>
            props>
        property>
    bean>
    <int-kafka:producer-context id="kafkaProducerContext"
        producer-properties="producerProperties">
        <int-kafka:producer-configurations>
            <int-kafka:producer-configuration broker-list="localhost:9092"
                       topic="test"
                       compression-codec="default"/>
        int-kafka:producer-configurations>
    int-kafka:producer-context>
beans>

int:channel是配置Spring Integration Channel, 此channel基于queue。
int-kafka:outbound-channel-adapter是outbound-channel-adapter对象， 内部使用一个线程池处理消息。关键是kafka-producer-context-ref。
int-kafka:producer-context配置producer列表，要处理的topic，这些Producer最终要转换成Kafka的Producer。

producer的配置参数如下：

broker-list             List of comma separated brokers that this producer connects to
topic                   Topic name or Java regex pattern of topic name
compression-codec       Compression method to be used. Default is no compression. Supported compression codec are gzip and snappy.
                        Anything else would result in no compression
value-encoder           Serializer to be used for encoding messages.
key-encoder             Serializer to be used for encoding the partition key
key-class-type          Type of the key class. This will be ignored if no key-encoder is provided
value-class-type        Type of the value class. This will be ignored if no value-encoder is provided.
partitioner             Custom implementation of a Kafka Partitioner interface.
async                   True/False - default is false. Setting this to true would make the Kafka producer to use
                        an async producer
batch-num-messages      Number of messages to batch at the producer. If async is false, then this has no effect.

value-encoder 和key-encoder可以是其它实现了Kafka Encoder接口的Bean。同样partitioner也是实现了Kafka的Partitioner接口的Bean。
一个Encoder的例子：

<bean id="kafkaEncoder" class="org.springframework.integration.kafka.serializer.avro.AvroSpecificDatumBackedKafkaEncoder">
    <constructor-arg value="com.company.AvroGeneratedSpecificRecord" />
bean>

Spring Integration Kafka 也提供了个基于Avro的Encoder。 Avro也是Apache的一个项目， 在大数据处理时也是一个常用的序列化框架。
不指定Encoder将使用Kafka缺省的Encoder (kafka.serializer.DefaultEncoder, byte[] –> same byte[])。

producerProperties可以用来设置配置属性进行调优。配置属性列表请参考 http://kafka.apache.org/documentation.html#producerconfigs

使用spring-integration-kafka接收消息

同样的原理实现一个消费者：

package com.colobu.spring_kafka_demo;
import java.util.Collection;
import java.util.HashMap;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.Random;
import java.util.Set;
import java.util.concurrent.ConcurrentHashMap;
import org.slf4j.LoggerFactory;
import org.springframework.context.support.ClassPathXmlApplicationContext;
import org.springframework.integration.channel.QueueChannel;
import org.springframework.messaging.Message;
import ch.qos.logback.classic.Level;
public class Consumer {
    private static final String CONFIG = "/consumer_context.xml";
    private static Random rand = new Random();
    @SuppressWarnings({ "unchecked", "unchecked", "rawtypes" })
    public static void main(String[] args) {
        ch.qos.logback.classic.Logger rootLogger = (ch.qos.logback.classic.Logger)LoggerFactory.getLogger(ch.qos.logback.classic.Logger.ROOT_LOGGER_NAME);
        rootLogger.setLevel(Level.toLevel("info"));

        final ClassPathXmlApplicationContext ctx = new ClassPathXmlApplicationContext(CONFIG, Consumer.class);
        ctx.start();
        final QueueChannel channel = ctx.getBean("inputFromKafka", QueueChannel.class);
        Message msg;        
        while((msg = channel.receive()) != null) {
            HashMap map = (HashMap)msg.getPayload();
            Set.Entry> set = map.entrySet();
            for (Map.Entry entry : set) {
                String topic = (String)entry.getKey();
                System.out.println("Topic:" + topic);
                ConcurrentHashMap> messages = (ConcurrentHashMap>)entry.getValue();
                Collection> values = messages.values();
                for (Iterator> iterator = values.iterator(); iterator.hasNext();) {
                    List list = iterator.next();
                    for (byte[] object : list) {
                        String message = new String(object);
                        System.out.println("\tMessage: " + message);
                    }

                }

            }

        }

        try {
            Thread.sleep(100000);
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
        ctx.close();
    }
}

Spring的配置文件如下：

<beans xmlns="http://www.springframework.org/schema/beans"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:int="http://www.springframework.org/schema/integration"
    xmlns:int-kafka="http://www.springframework.org/schema/integration/kafka"
    xmlns:task="http://www.springframework.org/schema/task"
    xsi:schemaLocation="http://www.springframework.org/schema/integration/kafka http://www.springframework.org/schema/integration/kafka/spring-integration-kafka.xsd
        http://www.springframework.org/schema/integration http://www.springframework.org/schema/integration/spring-integration.xsd
        http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd
        http://www.springframework.org/schema/task http://www.springframework.org/schema/task/spring-task.xsd">
    <int:channel id="inputFromKafka">
        <int:queue/>
    int:channel>

    <int-kafka:inbound-channel-adapter
        id="kafkaInboundChannelAdapter" kafka-consumer-context-ref="consumerContext"
        auto-startup="false" channel="inputFromKafka">
        <int:poller fixed-delay="10" time-unit="MILLISECONDS"
            max-messages-per-poll="5" />
    int-kafka:inbound-channel-adapter>
    <bean id="consumerProperties"
        class="org.springframework.beans.factory.config.PropertiesFactoryBean">
        <property name="properties">
            <props>
                <prop key="auto.offset.reset">smallestprop>
                <prop key="socket.receive.buffer.bytes">10485760prop> 
                <prop key="fetch.message.max.bytes">5242880prop>
                <prop key="auto.commit.interval.ms">1000prop>
            props>
        property>
    bean>
    <int-kafka:consumer-context id="consumerContext"
        consumer-timeout="4000" zookeeper-connect="zookeeperConnect" consumer-properties="consumerProperties">
        <int-kafka:consumer-configurations>
            <int-kafka:consumer-configuration
                group-id="mygroup" max-messages="5000">
                <int-kafka:topic id="test" streams="4" />
            int-kafka:consumer-configuration>
            
        int-kafka:consumer-configurations>
    int-kafka:consumer-context>
    <int-kafka:zookeeper-connect id="zookeeperConnect"
        zk-connect="localhost:2181" zk-connection-timeout="6000"
        zk-session-timeout="400" zk-sync-time="200" />
beans>

这个配置和Producer类似， 同样声明一个channel, 定义inbound-channel-adapter, 它引用Bean kafka-consumer-context,
kafka-consumer-context定义了消费者的列表。 consumer-configuration还提供了topic-filter，使用正则表达式建立白名单或者黑名单(exclude属性)。

消费者上下文还需要zookeeper-connect。

由于spring-integration-kafka只实现了high level Consumer API,这也就意味着你不可能回滚重新查看以前的消息， 因为high level API不提供offset管理。

注意Channel中得到的有效负载的类型是：
Map