千家信息网最后更新 2024年11月29日怎么联合使用Spark Streaming、Broadcast、Accumulaor
本篇内容介绍了"怎么联合使用Spark Streaming、Broadcast、Accumulaor"的有关知识,在实际案例的操作过程中,不少人都会遇到这样的困境,接下来就让小编带领大家学习一下如何处理这些情况吧!希望大家仔细阅读,能够学有所成!
广播可以自定义,通过Broadcast、Accumulator联合可以完成复杂的业务逻辑。
以下代码实现在本机9999端口监听,并向连接上的客户端发送单词,其中包含黑名单的单词Hadoop,Mahout和Hive。
package org.scala.opt
import java.io.{PrintWriter, IOException} import java.net.{Socket, SocketException, ServerSocket} case class ServerThread(socket : Socket) extends Thread("ServerThread") { override def run(): Unit = { val ptWriter = new PrintWriter(socket.getOutputStream) try { var count = 0 var totalCount = 0 var isThreadRunning : Boolean = true val batchCount = 1 val words = List("Java Scala C C++ C# Python JavaScript", "Hadoop Spark Ngix MFC Net Mahout Hive") while (isThreadRunning) { words.foreach(ptWriter.println) count += 1 if (count >= batchCount) { totalCount += count count = 0 println("batch " + batchCount + " totalCount => " + totalCount) Thread.sleep(1000) } //out.println此类中的方法不会抛出 I/O 异常,尽管其某些构造方法可能抛出异常。客户端可能会查询调用 checkError() 是否出现错误。 if(ptWriter.checkError()) { isThreadRunning = false println("ptWriter error then close socket") } } } catch { case e : SocketException => println("SocketException : ", e) case e : IOException => e.printStackTrace(); } finally { if (ptWriter != null) ptWriter.close() println("Client " + socket.getInetAddress + " disconnected") if (socket != null) socket.close() } println(Thread.currentThread().getName + " Exit") } } object SocketServer { def main(args : Array[String]) : Unit = { try { val listener = new ServerSocket(9999) println("Server is started, waiting for client connect...") while (true) { val socket = listener.accept() println("Client : " + socket.getLocalAddress + " connected") new ServerThread(socket).start() } listener.close() } catch { case e: IOException => System.err.println("Could not listen on port: 9999.") System.exit(-1) } } } |
以下代码实现接收本机9999端口发送的单词,统计黑名单出现的次数的功能。
package com.dt.spark.streaming_scala
import org.apache.spark.streaming.{Seconds, StreamingContext} import org.apache.spark.{SparkConf, Accumulator} import org.apache.spark.broadcast.Broadcast
/** * 第103课: 动手实战联合使用Spark Streaming、Broadcast、Accumulator实现在线黑名单过滤和计数 * 本期内容: 1,Spark Streaming与Broadcast、Accumulator联合 2,在线黑名单过滤和计算实战 */ object _103SparkStreamingBroadcastAccumulator {
@volatile private var broadcastList : Broadcast[List[String]] = null @volatile private var accumulator : Accumulator[Int] = null
def main(args : Array[String]) : Unit = { val conf = new SparkConf().setMaster("local[5]").setAppName("_103SparkStreamingBroadcastAccumulator") val ssc = new StreamingContext(conf, Seconds(5)) ssc.sparkContext.setLogLevel("WARN")
/** * 使用Broadcast广播黑名单到每个Executor中 */ broadcastList = ssc.sparkContext.broadcast(Array("Hadoop", "Mahout", "Hive").toList)
/** * 全局计数器,用于通知在线过滤了多少各黑名单 */ accumulator = ssc.sparkContext.accumulator(0, "OnlineBlackListCounter")
ssc.socketTextStream("localhost", 9999).flatMap(_.split(" ")).map((_, 1)).reduceByKey(_+_).foreachRDD {rdd =>{ if (!rdd.isEmpty()) { rdd.filter(wordPair => { if (broadcastList.value.contains(wordPair._1)) { println("BlackList word %s appeared".formatted(wordPair._1)) accumulator.add(wordPair._2) false } else { true } }).collect() println("BlackList appeared : %d times".format(accumulator.value)) } }} ssc.start() ssc.awaitTermination() ssc.stop() } } |
Server发送端日志如下,不断打印输出的次数。
Spark Streaming端打印黑名单的单词及出现的次数。
"怎么联合使用Spark Streaming、Broadcast、Accumulaor"的内容就介绍到这里了,感谢大家的阅读。如果想了解更多行业相关的知识可以关注网站,小编将为大家输出更多高质量的实用文章!