SparkStreaming数据源之Kafka

2,从Kafka里面拿数据的两种方式

(1) push(推过来的)

   kafka，flume  ->  Exeuctor内存-》磁盘  处理

    1）整个任务出问题了

    2）整个集群宕机了

    3）机房停电了

       数据有可能重复消费，也有可能漏了

 Spark 1.3 以后

 (2) pull(拉)

    sparkStreaming自己去维护消费到哪儿（HDFS上）0-50

   1）数据不会丢失

   2）数据不会重复消费

添加的依赖

<!-- https://mvnrepository.com/artifact/org.apache.spark/spark-streaming-kafka-0-8_2.11 -->         <dependency>             <groupId>org.apache.spark</groupId>             <artifactId>spark-streaming-kafka-0-8_2.11</artifactId>             <version>2.1.0</version>         </dependency>

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


object KafkaWordCountTest {
  def main(args: Array[String]): Unit= {
    val conf = new SparkConf().setAppName(s"${this.getClass.getSimpleName}").setMaster("local[2]")
    val ssc = new StreamingContext(conf,Seconds(2))

    //这个checkpoint一定要设置，因为streamingContext从kafka里面拿取数据的偏移量记录在这个hdfs的目录下面
    ssc.checkpoint("hdfs://hadoop1:9000/kafkastreaming")
    /**
      * ssc: StreamingContext,
      * kafkaParams: Map[String, String],
      * topics: Set[String]
      *
      */
    val kafkaParams = Map("metadata.broker.list" -> "hadoop1:9092")
    val topics=Set("htt")
    /**
      * K, V, KD <: Decoder[K], VD <: Decoder[V]
      *   k,v
      *
      *  k：位置信息
      *  v: 数据
      */
    val kafkaDStream = KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder](ssc, kafkaParams, topics)
      .map(_._2)

    kafkaDStream.flatMap(_.split(",")).map((_,1)).reduceByKey((_+_)).print()

    ssc.start()
    ssc.awaitTermination()
    ssc.stop()

  }

}