Spark Stream集成Flume实现词频统计（Pull方式）

使用Pull方式的优点

Spark Stream集成Flume有两种方式：

• Push-based Approach
• Pull-based Approach

那么我们在生产中该选择哪种方式呢？让我们来看看官网是怎么说的

由此可见，Pull相较于Push的可靠性和容错性更好，所以我们采用Pull的方式进行集成。

配置Flume

1. 配置pom.xml文件

   <scala.version>2.11.12scala.version>
   <spark.version>2.4.4spark.version>
   

   
   <dependency>
       <groupId>org.scala-langgroupId>
       <artifactId>scala-libraryartifactId>
       <version>${scala.version}version>
       
   dependency>
   
   <dependency>
       <groupId>org.apache.sparkgroupId>
       <artifactId>spark-streaming_2.11artifactId>
       <version>${spark.version}version>
   dependency>
   
   
   <dependency>
       <groupId>org.apache.sparkgroupId>
       <artifactId>spark-streaming-flume_2.11artifactId>
       <version>${spark.version}version>
   dependency>
   
   <dependency>
       <groupId>org.apache.sparkgroupId>
       <artifactId>spark-streaming-flume-sink_2.11artifactId>
       <version>${spark.version}version>
   dependency>
   
   <dependency>
       <groupId>org.apache.commonsgroupId>
       <artifactId>commons-lang3artifactId>
       <version>3.5version>
   dependency>
   

2. 配置Flume文件，name=streaming-flume-integration-pull

   integration-agent.sources = netcat_source
   integration-agent.sinks = spark_sink
   integration-agent.channels = memory_channel
   
   integration-agent.sources.netcat_source.type = netcat
   integration-agent.sources.netcat_source.bind = hlsijx
   integration-agent.sources.netcat_source.port = 9999
   
   integration-agent.sinks.spark_sink.type = org.apache.spark.streaming.flume.sink.SparkSink
   integration-agent.sinks.spark_sink.hostname = hlsijx
   integration-agent.sinks.spark_sink.port = 11111
   
   integration-agent.channels.memory_channel.type = memory
   
   integration-agent.sources.netcat_source.channels = memory_channel
   integration-agent.sinks.spark_sink.channel = memory_channel
   

配置Spark应用

1. 创建Scala文件，名叫FlumePullWordCount

   package com.hlsijx.spark.stream.wordcount
   
   import org.apache.spark.SparkConf
   import org.apache.spark.streaming.flume.FlumeUtils
   import org.apache.spark.streaming.{Seconds, StreamingContext}
   
   object FlumePullWordCount {
   
     def main(args: Array[String]): Unit = {
   
       if (args.length != 2){
         System.err.print("Usage: FlumePullWordCount  ")
         System.exit(1)
       }
       val Array(hostname, port) = args
   
       val sparkConf = new SparkConf()
   
       val ssc = new StreamingContext(sparkConf, Seconds(5))
   
       val lines = FlumeUtils.createPollingStream(ssc, hostname, port.toInt)
   
       val wordCount = lines.map(x => new String(x.event.getBody.array()).trim)
                       .flatMap(_.split(" ")).map((_, 1)).reduceByKey(_ + _)
   
       wordCount.print()
   
       ssc.start()
   
       ssc.awaitTermination()
     }
   }
   

2. 打包项目：mvn clean package -DskipTests，并上传到服务器路径：/data/spark-2.4.4-bin-2.6.0-cdh5.15.1/lib

3. 启动Flume

   flume-ng agent \
   --conf $FLUME_HOME/conf \
   --conf-file $FLUME_HOME/conf/streaming-flume-integration-pull.conf \
   --name integration-agent \
   -Dflume.root.logger=INFO,console
   

4. 启动Spark应用

   bin/spark-submit \
   --class com.hlsijx.spark.stream.wordcount.FlumePullWordCount \
   --master local[2] \
   --packages org.apache.spark:spark-streaming-flume_2.11:2.4.4 \
   /data/spark-2.4.4-bin-2.6.0-cdh5.15.1/lib/spark-1.0.jar hlsijx 11111
   

测试

启动telnet：telnet hlsijx 9999，输入a a a b，即可看到实时输出结果。