spring-sparkstreaming-kafka10集成实现和疑难杂症解决

一.前期准备

1.开发环境

window7

eclipse

jdk1.8

2.linux环境

zookeeper-3.4.8

hadoop-2.6.4

spark-1.6.0

scala-2.10.6

kafka_2.10-0.10.1.0

各环境的安装和部署请自行准备。

二.疑难杂症

1. spark+scala+kafka版本要一致

2. org.apache.spark.SparkException: A master URL must be set in your configuration

原因：

SparkConf中未设置master。由于Master为cluster程序管理中心，负责接收Client提交的作业，管理Worker，并命令Worker启动Driver和Executor。

解决：

在SparkConf中设置master，本处设置为“local[2]”.

local(default) 在本地而非集群跑Spark作业，并且只有一个worker thread（所以，并事实上没有并行）
local[k] 在本地跑Spark Application，有k个worker thread
spark://HOST:PORT 连接到指定URL的standalone集群
mesos://HOST:PORT 连接到指定的Mesos集群
yarn 连接到默认的YARN集群。yarn集群在SPARK_HOME/conf/yarn-site.xml中指定

3. object not serializable (class: org.apache.kafka.clients.consumer.ConsumerRecord)

原因：

ConsumerRecord是Kafka10版本自带的类。该类为接受kafka消息的key-value类，而异常时该类未序列化。而程序中调用JavaRdd的collect方法，而该方法需要将数据加载到内存，需要进行序列化。

解决：

方法1：将获取数据集合的形式转化单个数据获取。

方法2：通过kryo序列化ConsumerRecord类

三.代码实现

1. sparkStreaming+kafka处理

package com.lm.spark;

import java.io.Serializable;
import java.util.Arrays;
import java.util.Collection;
import java.util.HashMap;
import java.util.List;
import java.util.Map;

import org.apache.kafka.clients.consumer.ConsumerRecord;
import org.apache.kafka.common.serialization.StringDeserializer;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.JavaRDD;
import org.apache.spark.api.java.function.VoidFunction;
import org.apache.spark.api.java.function.VoidFunction2;
import org.apache.spark.streaming.Durations;
import org.apache.spark.streaming.Time;
import org.apache.spark.streaming.api.java.JavaInputDStream;
import org.apache.spark.streaming.api.java.JavaStreamingContext;
import org.apache.spark.streaming.kafka010.ConsumerStrategies;
import org.apache.spark.streaming.kafka010.KafkaUtils;
import org.apache.spark.streaming.kafka010.LocationStrategies;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.stereotype.Component;

@Component
public class SparkStreamingKafka implements Serializable {

	/**
	 * 
	 */
	private static final long serialVersionUID = 1L;

	public static Logger LOGGER = LoggerFactory.getLogger(SparkStreamingKafka.class);

	@Value("${spark.appname}")
	private String appName;
	@Value("${spark.master}")
	private String master;
	@Value("${spark.seconds}")
	private long second;
	@Value("${kafka.metadata.broker.list}")
	private String metadataBrokerList;
	@Value("${kafka.auto.offset.reset}")
	private String autoOffsetReset;
	@Value("${kafka.topics}")
	private String kafkaTopics;
	@Value("${kafka.group.id}")
	private String kafkaGroupId;

	public void processSparkStreaming() throws InterruptedException {
		// 1.配置sparkconf,必须要配置master
		SparkConf conf = new SparkConf().setAppName(appName).setMaster(master);
		conf.set("spark.serializer", "org.apache.spark.serializer.KryoSerializer");
		conf.set("spark.kryo.registrator", "com.lm.kryo.MyRegistrator");

		// 2.根据sparkconf 创建JavaStreamingContext
		JavaStreamingContext jsc = new JavaStreamingContext(conf, Durations.seconds(second));

		// 3.配置kafka
		Map<String, Object> kafkaParams = new HashMap<>();
		kafkaParams.put("bootstrap.servers", metadataBrokerList);
		kafkaParams.put("key.deserializer", StringDeserializer.class);
		kafkaParams.put("value.deserializer", StringDeserializer.class);
		kafkaParams.put("group.id", kafkaGroupId);
		kafkaParams.put("auto.offset.reset", autoOffsetReset);
		kafkaParams.put("enable.auto.commit", false);

		// 4.kafka主题
		Collection<String> topics = Arrays.asList(kafkaTopics.split(","));

		// 5.创建SparkStreaming输入数据来源input Stream
		final JavaInputDStream<ConsumerRecord<String, String>> stream =
				KafkaUtils.createDirectStream(jsc, LocationStrategies.PreferConsistent(),
						ConsumerStrategies.<String, String> Subscribe(topics, kafkaParams));

		// 6.spark rdd转化和行动处理
		stream.foreachRDD(new VoidFunction2<JavaRDD<ConsumerRecord<String, String>>, Time>() {

			private static final long serialVersionUID = 1L;

			@Override
			public void call(JavaRDD<ConsumerRecord<String, String>> v1, Time v2) throws Exception {

				List<ConsumerRecord<String, String>> consumerRecords = v1.collect();

				System.out.println("获取消息:" + consumerRecords.size());

			}
		});

		// 6. 启动执行
		jsc.start();
		// 7. 等待执行停止，如有异常直接抛出并关闭
		jsc.awaitTermination();
	}
}

2. kryo序列化接收器

package com.lm.kryo;

import org.apache.kafka.clients.consumer.ConsumerRecord;
import org.apache.spark.serializer.KryoRegistrator;

import com.esotericsoftware.kryo.Kryo;

public class MyRegistrator implements KryoRegistrator {

	@Override
	public void registerClasses(Kryo arg0) {
		arg0.register(ConsumerRecord.class);
	}

}

四.代码地址

github:https://github.com/a123demi/spring-sparkstreaming-kafka-10

oschina:http://git.oschina.net/a123demi/spring-sparkstreaming-kafka-10