spark streaming集成kafka
Kakfa起初是由LinkedIn公司开发的一个分布式的消息系统，后成为Apache的一部分，它使用Scala编写，以可水平扩展和高吞吐率而被广泛使用。目前越来越多的开源分布式处理系统如Cloudera、Apache Storm、Spark等都支持与Kafka集成。
Spark streaming集成kafka是企业应用中最为常见的一种场景。
一、安装kafka
参考文档：
http://kafka.apache.org/quickstart#quickstart_createtopic
1、安装java
略
2、安装zookeeper集群
参考：http://www.cnblogs.com/wcwen1990/p/6652105.html
3、安装scala
略
4、安装kafka
下载kafka安装文件：
https://archive.apache.org/dist/kafka/0.8.2.1/kafka_2.10-0.8.2.1.tgz
解压kafka安装包：
# tar -zxvf kafka_2.10-0.8.2.1.tgz -C /opt/cdh-5.3.6/
# chown -R hadoop:hadoop /opt/cdh-5.3.6/kafka_2.10-0.8.2.1/
删除kafka libs/zookeeper jar包,拷贝自己安装集群zookeeper jar包到kafka libs目录下：
$ rm libs/zookeeper-3.4.6.jar –rf
$ cp /opt/cdh-5.3.6/zookeeper-3.4.5-cdh5.3.6/zookeeper-3.4.5-cdh5.3.6.jar libs/
5、定义kafka配置文件
5.1）定义server.properties：
host.name=chavin.king
log.dirs=/opt/cdh-5.3.6/kafka_2.10-0.8.2.1/kafka-logs
zookeeper.connect=chavin.king:2181
定义producer.properties：
metadata.broker.list=chavin.king:9092
定义consumer.properties：
zookeeper.connect=chavin.king:2181
5.2）启动kafka server
$ bin/kafka-server-start.sh config/server.properties
$ jps
14020 NameNode
57749 Jps
14776 QuorumPeerMain
57690 Kafka
14507 NodeManager
14235 ResourceManager
14093 DataNode
14686 JobHistoryServer
57663 ZooKeeperMain
[zk: localhost:2181(CONNECTED) 3] ls /
[controller, controller_epoch, brokers, zookeeper, admin, consumers, config, hbase]
5.3）创建一个topic
$ bin/kafka-topics.sh --create --zookeeper chavin.king:2181 --replication-factor 1 --partitions 1 --topic test
$ bin/kafka-topics.sh --list --zookeeper chavin.king:2181
5.4）创建一个生产者，产生数据
$ bin/kafka-console-producer.sh --broker-list chavin.king:9092 --topic test
5.5）创建一个消费者，消费数据
$ bin/kafka-console-consumer.sh --zookeeper chavin.king:2181 --topic test --from-beginning
在生产者shell窗口输入数据，在消费者窗口可以看到数据输出到界面上。
二、spark streaming与kafka集成
参考文档：http://spark.apache.org/docs/1.3.0/streaming-kafka-integration.html
一）准备工作
1、编译spark，获得集成kafka jar包：
参考文档：http://www.cnblogs.com/wcwen1990/p/7688027.html
说明：spark streaming集成flume或者kafka需要一些支持jar包，这些jar包在编译spark过程中会自动在external目录下生成相应的jar文件，因此，这里需要编译spark来获得这些jar包。
Spark streaming集成kafka主要需要：spark-streaming-kafka_2.10-1.3.0.jar包。
2、集成相关jar包
$ cp external/kafka/target/spark-streaming-kafka_2.10-1.3.0.jar /opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/
$ cp libs/kafka_2.10-0.8.2.1.jar libs/kafka-clients-0.8.2.1.jar libs/zkclient-0.3.jar libs/metrics-core-2.2.0.jar /opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/
[externalLibs]$ ls
kafka_2.10-0.8.2.1.jar
kafka-clients-0.8.2.1.jar
metrics-core-2.2.0.jar
spark-streaming-kafka_2.10-1.3.0.jar
zkclient-0.3.jar
二）集成方式1：Receiver-based Approach
1、编写spark streaming集成kafka的wordcount
import java.util.HashMap
import org.apache.spark._
import org.apache.spark.streaming._
import org.apache.spark.streaming.StreamingContext._
import org.apache.spark.streaming.kafka._
val ssc = new StreamingContext(sc, Seconds(5))
val topicMap = Map("test" -> 1)
// read data
val lines = KafkaUtils.createStream(ssc, "chavin.king:2181", "testWordCountGroup", topicMap).map(_._2)
val words = lines.flatMap(_.split(" "))
val wordCounts = words.map(x => (x, 1)).reduceByKey(_ + _)
wordCounts.print()
ssc.start() // Start the computation
ssc.awaitTermination() // Wait for the computation to terminate
2、spark-shell local模式启动，并运行步骤1程序
bin/spark-shell --master local[2] --jars \
/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/spark-streaming-kafka_2.10-1.3.0.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/kafka_2.10-0.8.2.1.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/kafka-clients-0.8.2.1.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/zkclient-0.3.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/metrics-core-2.2.0.jar
scala> import java.util.HashMap
import java.util.HashMap
scala> import org.apache.spark._
import org.apache.spark._
scala> import org.apache.spark.streaming._
import org.apache.spark.streaming._
scala> import org.apache.spark.streaming.StreamingContext._
import org.apache.spark.streaming.StreamingContext._
scala> import org.apache.spark.streaming.kafka._
import org.apache.spark.streaming.kafka._
scala> val ssc = new StreamingContext(sc, Seconds(5))
ssc: org.apache.spark.streaming.StreamingContext = org.apache.spark.streaming.StreamingContext@1a28f9a0
scala> val topicMap = Map("test" -> 1)
topicMap: scala.collection.immutable.Map[String,Int] = Map(test -> 1)
scala> val lines = KafkaUtils.createStream(ssc, "chavin.king:2181", "testWordCountGroup", topicMap).map(_._2)
lines: org.apache.spark.streaming.dstream.DStream[String] = org.apache.spark.streaming.dstream.MappedDStream@27267641
scala>
scala> val words = lines.flatMap(_.split(" "))
words: org.apache.spark.streaming.dstream.DStream[String] = org.apache.spark.streaming.dstream.FlatMappedDStream@169b0639
scala> val wordCounts = words.map(x => (x, 1)).reduceByKey(_ + _)
wordCounts: org.apache.spark.streaming.dstream.DStream[(String, Int)] = org.apache.spark.streaming.dstream.ShuffledDStream@14f2b1ba
scala> wordCounts.print()
scala> ssc.start()
scala>ssc.awaitTermination()
3、测试
在kafka生产者shell端输入：
hadoop oracle mysql mysql mysql
这是我们在kafka消费者端可以看到如下输出：
hadoop oracle mysql mysql mysql
同时在spark streaming端也可以看到如下输出：
-------------------------------------------
Time: 1500021590000 ms
-------------------------------------------
(mysql,3)
(oracle,1)
(hadoop,1)
三）集成方式2：Direct Approach (No Receivers)
1、编写spark streaming集成kafka的wordcount
import kafka.serializer.StringDecoder
import org.apache.spark._
import org.apache.spark.streaming._
import org.apache.spark.streaming.StreamingContext._
import org.apache.spark.streaming.kafka._
val ssc = new StreamingContext(sc, Seconds(5))
val kafkaParams = Map[String, String]("metadata.broker.list" -> "chavin.king:9092")
val topicsSet = Set("test")
// read data
val messages = KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder](ssc, kafkaParams, topicsSet)
val lines = messages.map(_._2)
val words = lines.flatMap(_.split(" "))
val wordCounts = words.map(x => (x, 1)).reduceByKey(_ + _)
wordCounts.print()
ssc.start() // Start the computation
ssc.awaitTermination() // Wait for the computation to terminate
2、spark-shell local模式启动，并运行步骤1程序
bin/spark-shell --master local[2] --jars \
/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/spark-streaming-kafka_2.10-1.3.0.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/kafka_2.10-0.8.2.1.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/kafka-clients-0.8.2.1.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/zkclient-0.3.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/metrics-core-2.2.0.jar
scala> import kafka.serializer.StringDecoder
import kafka.serializer.StringDecoder
scala> import org.apache.spark._
import org.apache.spark._
scala> import org.apache.spark.streaming._
import org.apache.spark.streaming._
scala> import org.apache.spark.streaming.StreamingContext._
import org.apache.spark.streaming.StreamingContext._
scala> import org.apache.spark.streaming.kafka._
import org.apache.spark.streaming.kafka._
scala>
scala> val ssc = new StreamingContext(sc, Seconds(5))
ssc: org.apache.spark.streaming.StreamingContext = org.apache.spark.streaming.StreamingContext@2d05daca
scala>
scala> val kafkaParams = Map[String, String]("metadata.broker.list" -> "chavin.king:9092")
kafkaParams: scala.collection.immutable.Map[String,String] = Map(metadata.broker.list -> chavin.king:9092)
scala> val topicsSet = Set("test")
topicsSet: scala.collection.immutable.Set[String] = Set(test)
scala>
scala> // read data
scala> val messages = KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder](ssc, kafkaParams, topicsSet)
17/07/14 16:59:31 INFO VerifiableProperties: Verifying properties
17/07/14 16:59:31 INFO VerifiableProperties: Property group.id is overridden to
17/07/14 16:59:31 INFO VerifiableProperties: Property zookeeper.connect is overridden to
messages: org.apache.spark.streaming.dstream.InputDStream[(String, String)] = org.apache.spark.streaming.kafka.DirectKafkaInputDStream@375c2870
scala>
scala> val lines = messages.map(_._2)
lines: org.apache.spark.streaming.dstream.DStream[String] = org.apache.spark.streaming.dstream.MappedDStream@1dda179e
scala> val words = lines.flatMap(_.split(" "))
words: org.apache.spark.streaming.dstream.DStream[String] = org.apache.spark.streaming.dstream.FlatMappedDStream@996294c
scala> val wordCounts = words.map(x => (x, 1)).reduceByKey(_ + _)
wordCounts: org.apache.spark.streaming.dstream.DStream[(String, Int)] = org.apache.spark.streaming.dstream.ShuffledDStream@19cd9e6a
scala> wordCounts.print()
scala> ssc.start()
scala>ssc.awaitTermination()
3、测试
在kafka生产者shell端输入：
hadoop oracle mysql mysql mysql
这是我们在kafka消费者端可以看到如下输出：
hadoop oracle mysql mysql mysql
同时在spark streaming端也可以看到如下输出：
-------------------------------------------
Time: 1500021590000 ms
-------------------------------------------
(mysql,3)
(oracle,1)
(hadoop,1)
至此，spark streaming集成kafka两种方式演示OK。但是通过上述案例我们可以发现，目前的spark streaming仅仅对每次的输入值进行一次计算，而企业应用中，可能更需要将多次的输入值进行累加，那么该怎么实现呢？看下面案例？
四）使用UpdataStateByKey实现spark streaming多次输入值的累加操作
1、创建文件udsb.scala文件，输入如下内容：
$ cat udsb.scala
import kafka.serializer.StringDecoder
import org.apache.spark._
import org.apache.spark.streaming._
import org.apache.spark.streaming.StreamingContext._
import org.apache.spark.streaming.kafka._
val ssc = new StreamingContext(sc, Seconds(5))
ssc.checkpoint(".")
val kafkaParams = Map[String, String]("metadata.broker.list" -> "chavin.king:9092")
val topicsSet = Set("test")
val updateFunc = (values: Seq[Int], state: Option[Int]) => {
val currentCount = values.sum
val previousCount = state.getOrElse(0)
Some(currentCount + previousCount)
}
// read data
val messages = KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder](ssc, kafkaParams, topicsSet)
val lines = messages.map(_._2)
val words = lines.flatMap(_.split(" "))
val wordDstream = words.map(x => (x, 1))
val stateDstream = wordDstream.updateStateByKey[Int](updateFunc)
stateDstream.print()
ssc.start()
ssc.awaitTermination()
2、spark-shell local模式启动，并运行步骤1程序
bin/spark-shell --master local[2] --jars \
/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/spark-streaming-kafka_2.10-1.3.0.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/kafka_2.10-0.8.2.1.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/kafka-clients-0.8.2.1.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/zkclient-0.3.jar,/opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externalLibs/metrics-core-2.2.0.jar
scala> :load /opt/cdh-5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/udsb.scala
3、测试
在kafka生产者shell端输入：
3.1）第一次输入：hadoop oracle mysql
Spark streaming端可以看到如下输出：
-------------------------------------------
Time: 1500023985000 ms
-------------------------------------------
(mysql,1)
(oracle,1)
(hadoop,1)
3.2）第二次输入：hadoop oracle mysql
Spark streaming端可以看到如下输出：
-------------------------------------------
Time: 1500023985000 ms
-------------------------------------------
(mysql,2)
(oracle,2)
(hadoop,2)
3.3）第三次输入：hadoop oracle mysql
Spark streaming端可以看到如下输出：
-------------------------------------------
Time: 1500023985000 ms
-------------------------------------------
(mysql,3)
(oracle,3)
(hadoop,3)
spark streaming集成kafka的更多相关文章
- spark streaming集成kafka接收数据的方式
spark streaming是以batch的方式来消费,strom是准实时一条一条的消费.当然也可以使用trident和tick的方式来实现batch消费(官方叫做mini batch).效率嘛,有 ...
- 解决spark streaming集成kafka时只能读topic的其中一个分区数据的问题
1. 问题描述 我创建了一个名称为myTest的topic,该topic有三个分区,在我的应用中spark streaming以direct方式连接kakfa,但是发现只能消费一个分区的数据,多次更换 ...
- Spark Streaming与Kafka集成
Spark Streaming与Kafka集成 1.介绍 kafka是一个发布订阅消息系统,具有分布式.分区化.多副本提交日志特点.kafka项目在0.8和0.10之间引入了一种新型消费者API,注意 ...
- Spark Streaming之四：Spark Streaming 与 Kafka 集成分析
前言 Spark Streaming 诞生于2013年,成为Spark平台上流式处理的解决方案,同时也给大家提供除Storm 以外的另一个选择.这篇内容主要介绍Spark Streaming 数据接收 ...
- spark streaming集成flume
1. 安装flume flume安装,解压后修改flume_env.sh配置文件,指定java_home即可. cp hdfs jar包到flume lib目录下(否则无法抽取数据到hdfs上): $ ...
- Spark Streaming on Kafka解析和安装实战
本课分2部分讲解: 第一部分,讲解Kafka的概念.架构和用例场景: 第二部分,讲解Kafka的安装和实战. 由于时间关系,今天的课程只讲到如何用官网的例子验证Kafka的安装是否成功.后续课程会接着 ...
- spark streaming 对接kafka记录
spark streaming 对接kafka 有两种方式: 参考: http://group.jobbole.com/15559/ http://blog.csdn.net/kwu_ganymede ...
- Spark Streaming、Kafka结合Spark JDBC External DataSouces处理案例
场景:使用Spark Streaming接收Kafka发送过来的数据与关系型数据库中的表进行相关的查询操作: Kafka发送过来的数据格式为:id.name.cityId,分隔符为tab zhangs ...
- 【转】Spark Streaming和Kafka整合开发指南
基于Receivers的方法 这个方法使用了Receivers来接收数据.Receivers的实现使用到Kafka高层次的消费者API.对于所有的Receivers,接收到的数据将会保存在Spark ...
随机推荐
- springcloud如何实现服务的平滑发布
在之前的文章中我们提到服务的优雅下线,见: SpringCloud服务如何在Eureka安全优雅的下线 但这个对于ribbon调用其实是不平滑的,shutdown请求到后服务就马上关闭了,服务消费此时 ...
- HDU 1022.Train Problem I【栈的应用】【8月19】
Train Problem I Problem Description As the new term comes, the Ignatius Train Station is very busy n ...
- 全面理解Java内存模型(JMM)及volatile关键字(转载)
关联文章: 深入理解Java类型信息(Class对象)与反射机制 深入理解Java枚举类型(enum) 深入理解Java注解类型(@Annotation) 深入理解Java类加载器(ClassLoad ...
- 《Effective Java 第三版》目录汇总
经过反复不断的拖延和坚持,所有条目已经翻译完成,供大家分享学习.时间有限,个别地方翻译得比较仓促,希望有疑虑的地方指出批评改正. 第一章简介 忽略 第二章 创建和销毁对象 1. 考虑使用静态工厂方法替 ...
- 物联网架构成长之路(25)-Docker构建项目用到的镜像1
0. 前言 现在项目处于初级阶段,按照规划,先构建几个以后可能会用到的Image,并上传到阿里云的Docker仓库.以后博客中用到的Image,大部分都会用到这几个基础的Image,构建一个简单的物联 ...
- git忽略.idan目录
git rm -r --cached .idea git add . git commit -m '忽略idea' git pull git push
- 运行yarn的时候提示 node不是内部或外部命令
背景:准备react native 搭建,装完nodejs npm 重启cmd,再次管理员运行即可!
- 解决wps/office 1-2自动转换1月2日，用样式解决此问题
添加样式: td{mso-number-format:"\@"; }
- SparkR-Install
SparkR-Install 时间:2017-03-30 23:05:18 阅读:17 评论:0 收藏:0 [点我收藏+] 标签:too 下载 安装jd ...
- 当我们用自定义的类对象作为key时，我们必须在程序中覆盖HashCode()和equals()。
Key值既然可以作为对象,那么也可以用一个自定义的类.比如: m.put(new sutdent(“Liucy”,30),”boss”) 如果没有语句来判定Student类对象是否相同,则会全部打印出 ...