Spark Dstream 创建

3.Dstream 创建

　　Spark Streaming 原生支持一些不同的数据源。一些“核心”数据源已经被打包到 Spark

Streaming 的 Maven 工件中，而其他的一些则可以通过 spark-streaming-kafka 等附加工件获取。

每个接收器都以 Spark 执行器程序中一个长期运行的任务的形式运行，因此会占据分配给应用

的 CPU 核心。此外，我们还需要有可用的 CPU 核心来处理数据。这意味着如果要运行多个接

收器，就必须至少有和接收器数目相同的核心数，还要加上用来完成计算所需要的核心数。例如，

如果我们想要在流计算应用中运行 10 个接收器，那么至少需要为应用分配 11 个 CPU 核心。

所以如果在本地模式运行，不要使用 local 或者 local[1]。

 

 

 

3.1 文件数据源

3.1.1 用法及说明

　　文件数据流：能够读取所有 HDFS API 兼容的文件系统文件，通过 fileStream 方法进行读取，

Spark Streaming 将会监控 dataDirectory 目录并不断处理移动进来的文件，记住目前不支持嵌套目

录。

streamingContext.textFileStream(dataDirectory)

注意事项：

　　1）文件需要有相同的数据格式；

　　2）文件进入 dataDirectory 的方式需要通过移动或者重命名来实现；

　　3）一旦文件移动进目录，则不能再修改，即便修改了也不会读取新数据；

 

 

 

3.1.2 案例实操

（1）在 HDFS 上建好目录

[lxl@hadoop102 spark]$ hadoop fs -mkdir /fileStream

（2）在/opt/module/data 创建三个文件

[lxl@hadoop102 data]$ touch a.tsv
[lxl@hadoop102 data]$ touch b.tsv
[lxl@hadoop102 data]$ touch c.tsv

添加如下数据：
Helloatguigu
Hellospark

（3）编写代码

package com.lxl
import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}
import org.apache.spark.streaming.dstream.DStream
object FileStream {
  def main(args: Array[String]): Unit = {
    //1.初始化 Spark 配置信息
    val sparkConf = new SparkConf().setMaster("local[*]")
      .setAppName("StreamWordCount")
    //2.初始化 SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(5))
    //3.监控文件夹创建 DStream
    val dirStream = ssc.textFileStream("hdfs://hadoop102:9000/fileStream")
    //4.将每一行数据做切分，形成一个个单词
    val wordStreams = dirStream.flatMap(_.split("	"))
    //5.将单词映射成元组（word,1）
    val wordAndOneStreams = wordStreams.map((_, 1))
    //6.将相同的单词次数做统计
    val wordAndCountStreams = wordAndOneStreams.reduceByKey(_ + _)
    //7.打印
    wordAndCountStreams.print()
    //8.启动 SparkStreamingContext
    ssc.start()
    ssc.awaitTermination()
  }
}

（4）启动程序并向 fileStream 目录上传文件

[lxl@hadoop102 data]$ hadoop fs -put ./a.tsv /fileStream
[lxl@hadoop102 data]$ hadoop fs -put ./b.tsv /fileStream
[lxl@hadoop102 data]$ hadoop fs -put ./c.tsv /fileStream

（5）获取计算结果

-------------------------------------------
Time: 1539073810000 ms
-------------------------------------------
-------------------------------------------
Time: 1539073815000 ms
-------------------------------------------
(Hello,4)
(spark,2)
(atguigu,2)
-------------------------------------------
Time: 1539073820000 ms
-------------------------------------------
(Hello,2)
(spark,1)
(atguigu,1)
-------------------------------------------
Time: 1539073825000 ms
-------------------------------------------

3.2 RDD 队列

3.2.1 用法及说明

测试过程中，可以通过使用 ssc.queueStream(queueOfRDDs)来创建 DStream，每一个推送到

这个队列中的 RDD，都会作为一个 DStream 处理。

 

 

3.2.2 案例实操

1）需求：循环创建几个 RDD，将 RDD 放入队列。通过 SparkStream 创建 Dstream，计算 WordCount

 

2）编写代码

package com.atguigu
import org.apache.spark.SparkConf
import org.apache.spark.rdd.RDD
import org.apache.spark.streaming.dstream.{DStream, InputDStream}
import org.apache.spark.streaming.{Seconds, StreamingContext}
import scala.collection.mutable
object RDDStream {
  def main(args: Array[String]) {
    //1.初始化 Spark 配置信息
    val conf = new SparkConf().setMaster("local[*]").setAppName("RDDStream")
    //2.初始化 SparkStreamingContext
    val ssc = new StreamingContext(conf, Seconds(4))
    //3.创建 RDD 队列
    val rddQueue = new mutable.Queue[RDD[Int]]()
    //4.创建 QueueInputDStream
    val inputStream = ssc.queueStream(rddQueue,oneAtATime = false)
    //5.处理队列中的 RDD 数据
    val mappedStream = inputStream.map((_,1))
    val reducedStream = mappedStream.reduceByKey(_ + _)
    //6.打印结果
    reducedStream.print()
    //7.启动任务
    ssc.start()
    //8.循环创建并向 RDD 队列中放入 RDD
    for (i <- 1 to 5) {
      rddQueue += ssc.sparkContext.makeRDD(1 to 300, 10)
      Thread.sleep(2000)
    }
    ssc.awaitTermination()
  }
}

3）结果展示

-------------------------------------------
Time: 1539075280000 ms
-------------------------------------------
(4,60)
(0,60)
(6,60)
(8,60)
(2,60)
(1,60)
(3,60)
(7,60)
(9,60)
(5,60)
-------------------------------------------
Time: 1539075284000 ms
-------------------------------------------
(4,60)
(0,60)
(6,60)
(8,60)
(2,60)
(1,60)
(3,60)
(7,60)
(9,60)
(5,60)
-------------------------------------------
Time: 1539075288000 ms
-------------------------------------------
(4,30)
(0,30)
(6,30)
(8,30)
(2,30)
(1,30)
(3,30)
(7,30)
(9,30)
(5,30)
-------------------------------------------
Time: 1539075292000 ms
-------------------------------------------

3.3 自定义数据源

3.3.1 用法及说明

需要继承 Receiver，并实现 onStart、onStop 方法来自定义数据源采集。

 

3.3.2 案例实操

1）需求：自定义数据源，实现监控某个端口号，获取该端口号内容。

2）自定义数据源

 

package com.lxl

import java.io.{BufferedReader, InputStreamReader}
import java.net.Socket
import java.nio.charset.StandardCharsets
import org.apache.spark.storage.StorageLevel
import org.apache.spark.streaming.receiver.Receiver

class CustomerReceiver(host: String, port: Int) extends
  Receiver[String](StorageLevel.MEMORY_ONLY) {
  //最初启动的时候，调用该方法，作用为：读数据并将数据发送给 Spark
  override def onStart(): Unit = {
    new Thread("Socket Receiver") {
      override def run() {
        receive()
      }
    }.start()
  }
  //读数据并将数据发送给 Spark
  def receive(): Unit = {
    //创建一个 Socket
    var socket: Socket = new Socket(host, port)
    //定义一个变量，用来接收端口传过来的数据
    var input: String = null
    //创建一个 BufferedReader 用于读取端口传来的数据
    val reader = new BufferedReader(new InputStreamReader(socket.getInputStream,
      StandardCharsets.UTF_8))
    //读取数据
    input = reader.readLine()
    //当 receiver 没有关闭并且输入数据不为空，则循环发送数据给 Spark
    while (!isStopped() && input != null) {
      store(input)
      input = reader.readLine()
    }
    //跳出循环则关闭资源
    reader.close()
    socket.close()
    //重启任务
    restart("restart")
  }
  override def onStop(): Unit = {}
}
 

 

3）使用自定义的数据源采集数据

package com.atguigu

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}
import org.apache.spark.streaming.dstream.DStream

object FileStream {
  def main(args: Array[String]): Unit = {
    //1.初始化 Spark 配置信息
    Val sparkConf = new SparkConf().setMaster("local[*]").setAppName("StreamWordCount")

    //2.初始化 SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(5))

    //3.创建自定义 receiver 的 Streaming
    val lineStream = ssc.receiverStream(new CustomerReceiver("hadoop102", 9999))

    //4.将每一行数据做切分，形成一个个单词
    val wordStreams = lineStream.flatMap(_.split("	"))

    //5.将单词映射成元组（word,1）
    val wordAndOneStreams = wordStreams.map((_, 1))

    //6.将相同的单词次数做统计
    val wordAndCount = wordAndOneStreams.reduceByKey(_ + _)

    //7.打印
    wordAndCountStreams.print()

    //8.启动 SparkStreamingContext
    ssc.start()
    ssc.awaitTermination()
  }
}

 

3.4 Kafka 数据源

3.4.1 用法及说明

　　在工程中需要引入 Maven 工件 spark- streaming-kafka_2.10 来使用它。包内提供的

KafkaUtils 对象可以在 StreamingContext 和 JavaStreamingContext 中以你的 Kafka 消息创建出

DStream。由于 KafkaUtils 可以订阅多个主题，因此它创建出的 DStream 由成对的主题和消息

组成。要创建出一个流数据，需要使用 StreamingContext 实例、一个由逗号隔开的 ZooKeeper

主机列表字符串、消费者组的名字(唯一名字)，以及一个从主题到针对这个主题的接收器线程数

的映射表来调用 createStream() 方法。

 

 

3.4.2 案例实操

1）需求 1：通过 SparkStreaming 从 Kafka 读取数据，并将读取过来的数据做简单计

算(WordCount)，最终打印到控制台。

（1）导入依赖

        <!-- https://mvnrepository.com/artifact/org.apache.spark/spark-streaming-kafka -->
        <dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-streaming-kafka_2.11</artifactId>
            <version>1.6.3</version>
        </dependency>
        <!-- https://mvnrepository.com/artifact/org.apache.kafka/kafka-clients -->
        <dependency>
            <groupId>org.apache.kafka</groupId>
            <artifactId>kafka-clients</artifactId>
            <version>0.10.2.1</version>
        </dependency>

（2）编写代码

package com.lxl

import kafka.serializer.StringDecoder
import org.apache.kafka.clients.consumer.ConsumerConfig
import org.apache.spark.SparkConf
import org.apache.spark.rdd.RDD
import org.apache.spark.storage.StorageLevel
import org.apache.spark.streaming.dstream.ReceiverInputDStream
import org.apache.spark.streaming.kafka.KafkaUtils
import org.apache.spark.streaming.{Seconds, StreamingContext}

object KafkaSparkStreaming {
  def main(args: Array[String]): Unit = {

    //1.创建 SparkConf 并初始化 SSC
    val sparkConf: SparkConf = new SparkConf().setMaster("local[*]").setAppName("KafkaSparkStreaming")
    val ssc = new StreamingContext(sparkConf, Seconds(5))

    //2.定义 kafka 参数
    val zookeeper = "hadoop102:2181,hadoop103:2181,hadoop104:2181"
    val topic = "source"
    val consumerGroup = "spark"

    //3.将 kafka 参数映射为 map
    val kafkaParam: Map[String, String] = Map[String, String](
      ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG ->
        "org.apache.kafka.common.serialization.StringDeserializer",
      ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG ->
        "org.apache.kafka.common.serialization.StringDeserializer",
      ConsumerConfig.GROUP_ID_CONFIG -> consumerGroup,
      "zookeeper.connect" -> zookeeper
    )

    //4.通过 KafkaUtil 创建 kafkaDSteam
    val kafkaDSteam: ReceiverInputDStream[(String, String)] = KafkaUtils.createStream[String, String, StringDecoder, StringDecoder](
      ssc,
      kafkaParam,
      Map[String, Int](topic -> 3),
      StorageLevel.MEMORY_ONLY
    )

    //5.对 kafkaDSteam 做计算（WordCount）
    kafkaDSteam.foreachRDD {
      rdd => {
        val word: RDD[String] = rdd.flatMap(_._2.split(" "))
        val wordAndOne: RDD[(String, Int)] = word.map((_, 1))
        val wordAndCount: RDD[(String, Int)] = wordAndOne.reduceByKey(_ + _)
        wordAndCount.collect().foreach(println)
      }
    }

    //6.启动 SparkStreaming
    ssc.start()
    ssc.awaitTermination()
  }
}

笔记：

//启动kafka
[lxl@hadoop102 ~]$ /opt/module/kafka/bin/kafka-server-start.sh /opt/module/kafka/config/server.properties

[lxl@hadoop103 ~]$ /opt/module/kafka/bin/kafka-server-start.sh /opt/module/kafka/config/server.properties

[lxl@hadoop104 ~]$ /opt/module/kafka/bin/kafka-server-start.sh /opt/module/kafka/config/server.properties

//创建topic    *source
[lxl@hadoop102 ~]$ /opt/module/kafka/bin/kafka-topics.sh --zookeeper hadoop102:2181 --create --replication-factor 1 --partitions 2 --topic source

//启动生产者
[lxl@hadoop102 ~]$ /opt/module/kafka/bin/kafka-console-producer.sh --broker-list hadoop102:9092 --topic source

//创建topic    *target
[lxl@hadoop102 ~]$ /opt/module/kafka/bin/kafka-topics.sh --zookeeper hadoop102:2181 --create --replication-factor 1 --partitions 2 --topic target

//启动消费者
[lxl@hadoop102 ~]$ /opt/module/kafka/bin/kafka-console-consumer.sh --zookeeper hadoop102:2181 --from-beginning --topic target

package com.atlxl.kafkaStreaming

import java.util.Properties

import org.apache.commons.pool2.impl.{DefaultPooledObject, GenericObjectPool}
import org.apache.commons.pool2.{BasePooledObjectFactory, PooledObject}
import org.apache.kafka.clients.producer.{KafkaProducer, ProducerConfig, ProducerRecord}

class KafkaProxy(brokers:String){

  //存放配置文件
  private val pros:Properties = new Properties()
  pros.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG,brokers)
  pros.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringSerializer")
  pros.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringSerializer")

  val kafkaConn = new KafkaProducer[String,String](pros)

  def send(topic:String,key:String,value:String): Unit ={
    kafkaConn.send(new ProducerRecord[String,String](topic,key,value))
  }

  def send(topic:String,value:String): Unit ={
    kafkaConn.send(new ProducerRecord[String,String](topic,value))
  }

  def close: Unit ={
    kafkaConn.close()
  }

}

class KafkaProxyFactory(brokers:String) extends BasePooledObjectFactory[KafkaProxy]{

  //创建实例
  override def create(): KafkaProxy = new KafkaProxy(brokers)

  //将池中对象封装
  override def wrap(t: KafkaProxy): PooledObject[KafkaProxy] = new DefaultPooledObject[KafkaProxy](t)

}

object KafkaPool {

  //声明一个连接池对象
  var kafkaPool: GenericObjectPool[KafkaProxy] = null

  //
  def apply(brokers:String): GenericObjectPool[KafkaProxy] ={
    if (kafkaPool == null){
      KafkaPool.synchronized{
        if (kafkaPool == null){
          kafkaPool = new GenericObjectPool[KafkaProxy](new KafkaProxyFactory(brokers))
        }
      }
    }

    kafkaPool

  }


}

package com.atlxl.kafkaStreaming

import kafka.serializer.StringDecoder
import org.apache.kafka.clients.consumer.ConsumerConfig
import org.apache.spark.SparkConf
import org.apache.spark.streaming.kafka.KafkaUtils
import org.apache.spark.streaming.{Seconds, StreamingContext}

object KafkaStreaming {

  def main(args: Array[String]): Unit = {

    //conf
    val conf = new SparkConf().setAppName("kafka").setMaster("local[*]")
    val ssc = new StreamingContext(conf,Seconds(5))

    //kafka的参数
    val brokers = "hadoop102:9092"
    val zookeeper = "hadoop102:2181,hadoop103:2181,hadoop104:2181"
    val sourceTopic = "source"
    val targetTopic = "target"
    val consumerGroup = "consumer01"

    //封装kafka参数
    val kafkaParams = Map[String,String](
      ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG -> brokers,
      ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG -> "org.apache.kafka.common.serialization.StringDeserializer",
      ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG -> "org.apache.kafka.common.serialization.StringDeserializer",
      ConsumerConfig.GROUP_ID_CONFIG -> consumerGroup
    )


    val kafkaDStrem = KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder](ssc,kafkaParams,Set(sourceTopic))


    kafkaDStrem.foreachRDD{rdd =>
      rdd.foreachPartition{rddPar =>

        //创建生产者
        val kafkaPool = KafkaPool(brokers)
        val kafkaConn = kafkaPool.borrowObject()


        //写出到Kafka（targetTopic）
        //        val value = rddPar.map(x => x._2)
        for (item <- rddPar){
          //生产者发送数据
          kafkaConn.send(targetTopic,item._2)
        }

        //关闭生产者
        kafkaPool.returnObject(kafkaConn)
      }
    }


    /*//测试
    val result = kafkaDStrem.map(x => (x._1, x._2)).reduceByKey(_+_)
    result.print()*/

    ssc.start()
    ssc.awaitTermination()


  }

}