spark学习笔记（十二）——sparkStreaming-RDD队列/自定义数据源/kafka数据源/DStream转换/DStream输出/优雅关闭

目录

RDD队列

自定义数据源

 kafka数据源

DStream转换

无状态转化操作

有状态转化操作

DStream输出

优雅关闭

用到的全部依赖：

    <dependencies>
        <dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-core_2.12</artifactId>
            <version>3.0.0</version>
        </dependency>
 
        <dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-sql_2.12</artifactId>
            <version>3.0.0</version>
        </dependency>
 
        <dependency>
            <groupId>mysql</groupId>
            <artifactId>mysql-connector-java</artifactId>
            <version>5.1.27</version>
        </dependency>
 
        <dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-hive_2.12</artifactId>
            <version>3.0.0</version>
        </dependency>
 
        <dependency>
            <groupId>org.apache.hive</groupId>
            <artifactId>hive-exec</artifactId>
            <version>1.2.1</version>
        </dependency>
 
        <dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-streaming_2.12</artifactId>
            <version>3.0.0</version>
        </dependency>
 
        <dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-streaming-kafka-0-8_2.11</artifactId>
            <version>2.4.1</version>
        </dependency>
 
        <dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-streaming-kafka-0-10_2.12</artifactId>
            <version>3.0.0</version>
        </dependency>
 
        <dependency>
            <groupId>com.fasterxml.jackson.core</groupId>
            <artifactId>jackson-core</artifactId>
            <version>2.10.1</version>
        </dependency>
 
    </dependencies>
    <build>
        <plugins>
            <!--该插件用于把Scala代码编译成为class文件-->
            <plugin>
                <groupId>net.alchim31.maven</groupId>
                <artifactId>scala-maven-plugin</artifactId>
                <version>3.2.2</version>
                <executions>
                    <execution>
                        <!--声明绑定到maven的compile阶段-->
                        <goals>
                            <goal>testCompile</goal>
                        </goals>
                    </execution>
                </executions>
            </plugin>
            <plugin>
                <groupId>org.apache.maven.plugins</groupId>
                <artifactId>maven-assembly-plugin</artifactId>
                <version>3.1.0</version>
                <configuration>
                    <descriptorRefs>
                        <descriptorRef>jar-with-dependencies</descriptorRef>
                    </descriptorRefs>
                </configuration>
                <executions>
                    <execution>
                        <id>make-assembly</id>
                        <phase>package</phase>
                        <goals>
                            <goal>single</goal>
                        </goals>
                    </execution>
                </executions>
            </plugin>
        </plugins>
    </build>

RDD队列

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。

循环创建几个RDD并放入队列。通过SparkStream创建Dstream计算WordCount。

代码：

import org.apache.spark.SparkConf
import org.apache.spark.rdd.RDD
import org.apache.spark.streaming.{Seconds, StreamingContext}
import scala.collection.mutable
 
object RDD {
  def main(args: Array[String]): Unit = {
    //TODO 创建配置环境
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("RDD")
    val ssc = new StreamingContext(sparkConf,Seconds(5))
 
    //TODO 操作
    //创建RDD队列
    val Rdd = new mutable.Queue[RDD[Int]]()
    //创建queueStream
    val imputStream = ssc.queueStream(Rdd, oneAtATime = false)
    //处理RDD数据
    val mapRDD = imputStream.map((_, 1))
    val reduceRDD = mapRDD.reduceByKey(_ + _)
    //打印
    reduceRDD.print()
    //启动任务
    ssc.start()
    //循环创建RDD
    for (i <- 1 to 50) {
      Rdd += ssc.sparkContext.makeRDD(1 to 100,5)
      Thread.sleep(2000)
    }
 
    ssc.awaitTermination()
  }
 
}

结果：

自定义数据源

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

案例1代码：

import org.apache.spark.SparkConf
import org.apache.spark.storage.StorageLevel
import org.apache.spark.streaming.receiver.Receiver
import org.apache.spark.streaming.{Seconds, StreamingContext}
import scala.util.Random
 
object UserDefined_DataSource {
  def main(args: Array[String]): Unit = {
    //TODO 创建配置环境
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("RDD")
    val ssc = new StreamingContext(sparkConf,Seconds(5))
    //采集数据
    val messageDS = ssc.receiverStream(new MyReceiver())
    messageDS.print()
    //开始
    ssc.start()
    ssc.awaitTermination()
  }
  /*
  自定义数据采集器
  1.继承Receiver，定义泛型，传递参数
  2.重写方法
  */
  class MyReceiver extends Receiver[String](StorageLevel.MEMORY_ONLY) {
    private var fig = true
    //最初启动,读数据
    override def onStart(): Unit = {
      new Thread(new Runnable {
        override def run(): Unit = {
          while (true) {
            val message = "采集的数据为：" + new Random().nextInt(10).toString
            store(message)
            Thread.sleep(500)
          }
        }
      }).start()
    }
    //停止
    override def onStop(): Unit = {
      fig = false
    }
  }
}

结果：

案列2代码：

import java.io.{BufferedReader, InputStreamReader}
import java.net.Socket
import java.nio.charset.StandardCharsets
import org.apache.spark.SparkConf
import org.apache.spark.storage.StorageLevel
import org.apache.spark.streaming.receiver.Receiver
import org.apache.spark.streaming.{Seconds, StreamingContext}
 
object UserDefined_DataSource02 {
  def main(args: Array[String]): Unit = {
    //初始化Spark配置信息
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("WordCount")
    //初始化SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(10))
    //创建自定义数据接收器
    val lineStream = ssc.receiverStream(new MyReceiver("hadoop01", 9999))
    //数据切分
    val word = lineStream.flatMap(_.split("\t"))
    //映射为（word，1）
    val word2 = word.map((_, 1))
    //统计
    val wordCount = word2.reduceByKey(_ + _)
    //7.打印
    wordCount.print()
    //启动
    ssc.start()
    ssc.awaitTermination()
  }
  /*
  自定义数据采集器
  1.继承Receiver，定义泛型，传递参数
  2.重写方法
  */
  class MyReceiver(host: String,port: Int) extends Receiver[String](StorageLevel.MEMORY_ONLY) {
    //最初启动，调用方法
    override def onStart(): Unit = {
      new Thread("Socket Receiver") {
        override def run() {
          receive()
        }
      }.start()
    }
    //读数据并发送给Spark
    def receive(): Unit = {
      //创建Socket
      var socket: Socket = new Socket(host, port)
      //定义变量，接收数据
      var input: String = null
      //创建BufferedReader,读取数据
      val reader = new BufferedReader(new InputStreamReader(socket.getInputStream, StandardCharsets.UTF_8))
      //读数据
      input = reader.readLine()
      //如果receiver没有关闭且输入数据不为空
      //循环发送数据给Spark
      while (!isStopped() && input != null) {
        store(input)
        input = reader.readLine()
      }
      //跳出循环，关闭资源
      reader.close()
      socket.close()
      //重启
      restart("restart")
    }
    override def onStop(): Unit = {}
  }
}

 结果：

 kafka数据源

ReceiverAPI：需要一个专门的Executor接收数据，然后发送给其他的Executor计算。存在数据的节点内存溢出问题。

DirectAPI：由计算的Executor主动消费Kafka的数据，速度由自身控制。

添加依赖

        <dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-streaming-kafka-0-10_2.12</artifactId>
            <version>3.0.0</version>
        </dependency>
 
        <dependency>
            <groupId>com.fasterxml.jackson.core</groupId>
            <artifactId>jackson-core</artifactId>
            <version>2.10.1</version>
        </dependency>

代码：

import org.apache.kafka.clients.consumer.{ConsumerConfig, ConsumerRecord}
import org.apache.spark.SparkConf
import org.apache.spark.streaming.dstream.InputDStream
import org.apache.spark.streaming.kafka010.{ConsumerStrategies, KafkaUtils, LocationStrategies}
import org.apache.spark.streaming.{Seconds, StreamingContext}
 
object UserDefined_DataSource03_kafka {
  def main(args: Array[String]): Unit = {
    //初始化Spark配置信息
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("Kafka")
    //初始化SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(15))
    //定义kafka参数
    val kafkaPara: Map[String, Object] = Map[String, Object](
      ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG -> "hadoop01:9092",
      ConsumerConfig.GROUP_ID_CONFIG -> "testkafka",
      "key.deserializer" -> "org.apache.kafka.common.serialization.StringDeserializer",
      "value.deserializer" -> "org.apache.kafka.common.serialization.StringDeserializer"
    )
    //读取Kafka数据创建DStream
    val kafkaDStream: InputDStream[ConsumerRecord[String, String]] =
      KafkaUtils.createDirectStream[String, String](
        ssc,
        LocationStrategies.PreferConsistent,
        ConsumerStrategies.Subscribe[String, String](Set("testkafka"), kafkaPara))
    //取出KV对
    val value = kafkaDStream.map(record => record.value())
    //计算
    value.flatMap(_.split(" "))
        .map((_,1))
        .reduceByKey(_+_)
        .print()
    //启动
    ssc.start()
    ssc.awaitTermination()
  }
}

测试：

1）开启zookeeper

2）开启kafka

3）创建主题

进入kafka目录，运行：

bin/kafka-topics.sh --create --zookeeper hadoop01:2181 --replication-factor 1 --partitions 1 --topic testkafka

4）打开生产者

进入kafka目录，运行：

bin/kafka-console-producer.sh --broker-list hadoop01:9092 --topic testkafka

5）打开消费者

打开另一个窗口，进入kafka目录，运行：

bin/kafka-console-consumer.sh --bootstrap-server hadoop01:9092 --topic testkafka

6）运行程序

结果

DStream转换

DStream的操作和RDD类似，分为Transformations（转换）和Output Operations（输出）两种，转换操作中还有一些比较特殊的原语：updateStateByKey()、transform()以及各种Window相关的原语。

无状态转化和有状态转化：是否保存一个采集周期的数据，保存就是有状态，不保存就是无状态。

无状态转化操作

无状态转化操作就是把简单的RDD转化操作应用到每个批次上，也就是转化DStream中的每一个 RDD。

注意：1）针对键值对的DStream转化操作要添加import StreamingContext._才能在Scala上使用。

2）每个DStream在内部是由许多RDD组成，且无状态转化操作是分别应用到每个RDD上，例如：reduceByKey()会归约每个时间区间中的数据，但不会归约不同区间之间的数据。

join

两个流之间的join需要两个流的批次大小一致才能做到同时触发计算。计算过程就是对当前批次的两个流中各自的RDD进行join使与两个RDD的join效果相同。

代码：

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}
 
object Join {
  def main(args: Array[String]): Unit = {
 
    //初始化Spark配置信息
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("Join")
    //初始化SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(15) )
    //从端口获取数据创建流
    val line01 = ssc.socketTextStream("hadoop01", 9999)
    val line02 = ssc.socketTextStream("hadoop01", 8888)
    //转换为KV类型
    val word01 = line01.flatMap(_.split(" ")).map((_,1))
    val word02 = line02.flatMap(_.split(" "))map((_,"YES"))
    //JOIN
    val join = word01.join(word02)
    //打印
    join.print()
    //启动任务
    ssc.start()
    ssc.awaitTermination()
  }
}

测试结果：

transform

transform允许DStream上执行任意的RDD-to-RDD函数；该函数并没有在DStream的API中暴露出来方便扩展Spark API；该函数每一批次调度一次；其实也就是对DStream中的RDD应用转换。

 代码：

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}
 
object Transform {
  def main(args: Array[String]): Unit = {
 
    //初始化Spark配置信息
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("Traneform")
    //初始化SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(15) )
    //创建Dstream
    val line = ssc.socketTextStream("hadoop01", 9999)
    //转化为RDD
    val DStreamToRDD = line.transform(
      rdd => {
        val word = rdd.flatMap(_.split(" "))
        val wordmap = word.map((_, 1))
        val value = wordmap.reduceByKey(_ + _)
        value
      }
    )
    //打印
    DStreamToRDD.print()
    //启动任务
    ssc.start()
    ssc.awaitTermination()
  }
}

 测试结果：

有状态转化操作

UpdateStateByKey

UpdateStateByKey原语用于记录历史记录。

有时我们需要在DStream中跨批次维护状态(如：流计算中累加wordcount)；updateStateByKey()提供对一个状态变量的访问，给一个由(键，事件)对构成的DStream，传递一个函数，该函数指定根据新的事件更新每个键对应的状态并构建出一个新的DStream，其内部数据为(键，状态) 对。

updateStateByKey() 的结果是一个新的DStream，其内部的RDD序列是由每个时间区间对应的(键，状态)对组成的。

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}
 
object UpdateStateByKey {
  def main(args: Array[String]): Unit = {
 
    //初始化Spark配置信息
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("Count")
    //初始化SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(3))
    //使用有状态操作时需要设置检查点路径
    ssc.checkpoint("/ck")
    //创建DStream
    val lines = ssc.socketTextStream("hadoop01", 9999)
    val wordMap = lines.map((_,1))
    //updateStateByKey：根据key进行数据状态更新
    //第一个值：相同key的value数据
    //第二个值：缓冲区相同key的value数据
    val wordCount = wordMap.updateStateByKey(
      (seq: Seq[Int], buff: Option[Int]) => {
        val newCount = buff.getOrElse(0) + seq.sum
        Option(newCount)
      }
    )
    wordCount.print()
    //启动任务
    ssc.start()
    ssc.awaitTermination()
  }
 
}

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}
 
object WindowOperations {
  def main(args: Array[String]): Unit = {
    //TODO 采集数据周期为：3秒   窗口时长为：12秒    滑动周期为：6秒
 
    //初始化Spark配置信息
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("SparkStreaming")
    //初始化SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(3) )
    //创建Dstream
    val line = ssc.socketTextStream("hadoop01", 9999)
    //window（窗口时长，滑动周期）
    //窗口的范围应该是采集数据周期的整数倍
    //默认以一个采集周期进行滑动
    //为了避免重复数据可以改变滑动的幅度,窗口时长=滑动周期时避免重复计算
    val word = line.map((_, 1))
    //窗口时长为：12秒    滑动周期为：6秒
    val windowDS = word.window(Seconds(12),Seconds(6))
    val Count = windowDS.reduceByKey(_ + _)
    Count.print()
    //启动任务
    ssc.start()
    ssc.awaitTermination()
  }
}

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}
 
object WindowOperations2 {
  def main(args: Array[String]): Unit = {
    //TODO 采集数据周期为：3秒   窗口时长为：12秒    滑动周期为：6秒
 
    //初始化Spark配置信息
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("SparkStreaming")
    //初始化SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(3) )
    ssc.checkpoint("ck")
    //创建Dstream
    val line = ssc.socketTextStream("hadoop01", 9999)
    val word = line.map((_, 1))
    //窗口时长为：12秒    滑动周期为：6秒
    val windowDS = word.reduceByKeyAndWindow(
      (x:Int,y:Int) => {x + y},
      (x:Int,y:Int) => {x - y},
      Seconds(6),
      Seconds(3)
    )
    windowDS.print()
    //启动任务
    ssc.start()
    ssc.awaitTermination()
  }
}

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext, StreamingContextState}
 
object MonitorStop2 {
  def main(args: Array[String]): Unit = {
    /*
    线程关闭
    val thread = new Thread()
    thread.start()//线程开启
    thread.stop()//强制关闭
    */
 
    //初始化Spark配置信息
    val sparkConf = new SparkConf().setMaster("local[*]").setAppName("SparkStreaming")
    //初始化SparkStreamingContext
    val ssc = new StreamingContext(sparkConf, Seconds(3) )
    //创建Dstream
    val line = ssc.socketTextStream("hadoop01", 9999)
    val word = line.map((_, 1))
    word.print()
    //启动任务
    ssc.start()
    //创建新的线程用作关闭线程
    //而且需要在第三方程序中增加关闭程序
    new Thread(
      new Runnable {
        override def run(): Unit = {
          //优雅关闭：计算节点不再接收新的数据，把已经有的数据处理完毕后关闭
          //Mysql:Table(stopSpark) => Row => data
          //Redis:Data(K-V)
          //ZK:/stopSpark
          //HDFS:/stopSpark
            Thread.sleep(5000)
            //获取SparkStreaming状态
            val state = ssc.getState()
            if (state == StreamingContextState.ACTIVE) {
              ssc.stop(true,true)
          }
          //退出线程
          System.exit(0)
        }
      }
    ).start()
    ssc.awaitTermination()    //block阻塞main线程
  }
}

import java.net.URI
import org.apache.hadoop.conf.Configuration
import org.apache.hadoop.fs.{FileSystem, Path}
import org.apache.spark.streaming.{StreamingContext, StreamingContextState}
 
class MonitorStop(ssc: StreamingContext) extends Runnable {
  override def run(): Unit = {
    val fs: FileSystem = FileSystem.get(new URI("hdfs://hadoop01:9000"), new
        Configuration(), "spark")
    while (true) {
      try
        Thread.sleep(5000)
      catch {
        case e: InterruptedException =>
          e.printStackTrace()
      }
      val state: StreamingContextState = ssc.getState
      val bool: Boolean = fs.exists(new Path("hdfs://hadoop01:9000/stopSpark"))
      if (bool) {
        if (state == StreamingContextState.ACTIVE) {
          ssc.stop(stopSparkContext = true, stopGracefully = true)
          System.exit(0)
        }
      }
    }
  }
}

import org.apache.spark.SparkConf
import org.apache.spark.streaming.dstream.{DStream, ReceiverInputDStream}
import org.apache.spark.streaming.{Seconds, StreamingContext}
object SparkTest {
  def createSSC(): _root_.org.apache.spark.streaming.StreamingContext = {
    val update: (Seq[Int], Option[Int]) => Some[Int] = (values: Seq[Int], status:
    Option[Int]) => {
      //当前批次内容的计算
      val sum: Int = values.sum
      //取出状态信息中上一次状态
      val lastStatu: Int = status.getOrElse(0)
      Some(sum + lastStatu)
    }
    val sparkConf: SparkConf = new SparkConf().setMaster("local[*]").setAppName("SparkTest")
    //设置优雅的关闭
    sparkConf.set("spark.streaming.stopGracefullyOnShutdown", "true")
    val ssc = new StreamingContext(sparkConf, Seconds(5))
    ssc.checkpoint("./ck")
    val line: ReceiverInputDStream[String] = ssc.socketTextStream("hadoop01", 9999)
    val word: DStream[String] = line.flatMap(_.split(" "))
    val wordAndOne: DStream[(String, Int)] = word.map((_, 1))
    val wordAndCount: DStream[(String, Int)] = wordAndOne.updateStateByKey(update)
    wordAndCount.print()
    ssc
  }
  def main(args: Array[String]): Unit = {
    val ssc: StreamingContext = StreamingContext.getActiveOrCreate("./ck", () => createSSC())
    new Thread(new MonitorStop(ssc)).start()
    ssc.start()
    ssc.awaitTermination()
  }
}

恢复数据

设置checkpoint用于恢复数据。

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}
 
object RecoverData {
  def main(args: Array[String]): Unit = {
    //TODO 恢复数据
    val ssc = StreamingContext.getActiveOrCreate("ck", () => {
      //初始化Spark配置信息
      val sparkConf = new SparkConf().setMaster("local[*]").setAppName("SparkStreaming")
      //初始化SparkStreamingContext
      val ssc = new StreamingContext(sparkConf, Seconds(3))
      //创建Dstream
      val line = ssc.socketTextStream("hadoop01", 9999)
      val word = line.map((_, 1))
      word.print()
      ssc
    })
    ssc.checkpoint("ck")
    //启动任务
    ssc.start()
    ssc.awaitTermination()    //block阻塞main线程
  }
}

本文仅仅是学习笔记的记录！！！