分布式消息队列kafka系列介绍 — 核心API介绍及实例

一 Producer的API

1.Producer的创建,依赖于ProducerConfig
public Producer(ProducerConfig config);

2.单个或是批量的消息发送
public void send(KeyedMessage<K,V> message);
public void send(List<KeyedMessage<K,V>> messages);

3.关闭Producer到所有broker的连接
public void close();

二 Consumer的高层API

主要是Consumer和ConsumerConnector,这里的Consumer是ConsumerConnector的静态工厂类
class Consumer {
public static kafka.javaapi.consumer.ConsumerConnector createJavaConsumerConnector(config: ConsumerConfig);
}

具体的消息的消费都是在ConsumerConnector中
创建一个消息处理的流,包含所有的topic,并根据指定的Decoder
public <K,V> Map<String, List<KafkaStream<K,V>>>
createMessageStreams(Map<String, Integer> topicCountMap, Decoder<K> keyDecoder, Decoder<V> valueDecoder);

创建一个消息处理的流,包含所有的topic,使用默认的Decoder
public Map<String, List<KafkaStream<byte[], byte[]>>> createMessageStreams(Map<String, Integer> topicCountMap);

获取指定消息的topic,并根据指定的Decoder
public <K,V> List<KafkaStream<K,V>>
createMessageStreamsByFilter(TopicFilter topicFilter, int numStreams, Decoder<K> keyDecoder, Decoder<V> valueDecoder);

获取指定消息的topic,使用默认的Decoder
public List<KafkaStream<byte[], byte[]>> createMessageStreamsByFilter(TopicFilter topicFilter);

提交偏移量到这个消费者连接的topic
public void commitOffsets();

关闭消费者
public void shutdown();

高层的API中比较常用的就是public List<KafkaStream<byte[], byte[]>> createMessageStreamsByFilter(TopicFilter topicFilter);和public void commitOffsets();

三 Consumer的简单API–SimpleConsumer

批量获取消息
public FetchResponse fetch(request: kafka.javaapi.FetchRequest);

获取topic的元信息
public kafka.javaapi.TopicMetadataResponse send(request: kafka.javaapi.TopicMetadataRequest);

获取目前可用的偏移量
public kafka.javaapi.OffsetResponse getOffsetsBefore(request: OffsetRequest);

关闭连接
public void close();

对于大部分应用来说,高层API就已经足够使用了,但是若是想做更进一步的控制的话,可以使用简单的API,例如消费者重启的情况下,希望得到最新的offset,就该使用SimpleConsumer.

四 Kafka Hadoop Consumer API

提供了一个可水平伸缩的解决方案来结合hadoop的使用参见
https://github.com/linkedin/camus/tree/camus-kafka-0.8/

五 实战

maven依赖:

<dependency>
<groupId>org.apache.kafka</groupId>
<artifactId>kafka_2.10</artifactId>
<version>0.8.0</version>
</dependency>

生产者代码:


import kafka.javaapi.producer.Producer;
import kafka.producer.KeyedMessage;
import kafka.producer.ProducerConfig;

import java.util.Properties;

/**
 * <pre>
 * Created by zhaoming on 14-5-4 下午3:23
 * </pre>
 */
public class KafkaProductor {

public static void main(String[] args) throws InterruptedException {

Properties properties = new Properties();
 properties.put("zk.connect", "127.0.0.1:2181");
 properties.put("metadata.broker.list", "localhost:9092");

properties.put("serializer.class", "kafka.serializer.StringEncoder");

ProducerConfig producerConfig = new ProducerConfig(properties);
 Producer<String, String> producer = new Producer<String, String>(producerConfig);

// 构建消息体
 KeyedMessage<String, String> keyedMessage = new KeyedMessage<String, String>("test-topic", "test-message");
 producer.send(keyedMessage);

Thread.sleep(1000);

producer.close();
 }

}

消费端代码

import java.io.UnsupportedEncodingException;
import java.util.List;
import java.util.Properties;
import java.util.concurrent.TimeUnit;

import kafka.consumer.*;
import kafka.javaapi.consumer.ConsumerConnector;
import kafka.message.MessageAndMetadata;

import org.apache.commons.collections.CollectionUtils;

/**
 * <pre>
 * Created by zhaoming on 14-5-4 下午3:32
 * </pre>
 */
public class kafkaConsumer {

public static void main(String[] args) throws InterruptedException, UnsupportedEncodingException {

Properties properties = new Properties();
 properties.put("zookeeper.connect", "127.0.0.1:2181");
 properties.put("auto.commit.enable", "true");
 properties.put("auto.commit.interval.ms", "60000");
 properties.put("group.id", "test-group");

ConsumerConfig consumerConfig = new ConsumerConfig(properties);

ConsumerConnector javaConsumerConnector = Consumer.createJavaConsumerConnector(consumerConfig);

 //topic的过滤器
 Whitelist whitelist = new Whitelist("test-topic");
 List<KafkaStream<byte[], byte[]>> partitions = javaConsumerConnector.createMessageStreamsByFilter(whitelist);

if (CollectionUtils.isEmpty(partitions)) {
 System.out.println("empty!");
 TimeUnit.SECONDS.sleep(1);
 }

//消费消息
 for (KafkaStream<byte[], byte[]> partition : partitions) {

ConsumerIterator<byte[], byte[]> iterator = partition.iterator();
 while (iterator.hasNext()) {
 MessageAndMetadata<byte[], byte[]> next = iterator.next();
 System.out.println("partiton:" + next.partition());
 System.out.println("offset:" + next.offset());
 System.out.println("message:" + new String(next.message(), "utf-8"));
 }

}

}
}

PS:感觉消费端的API设计实在太难用了。

作者: inter12

在这苦短的人生中,追求点自己的简单快乐

发表评论

电子邮件地址不会被公开。 必填项已用*标注