如何在spark-streaming中获取通过kafka传递的flume信息header

标题好长。。。好复杂。。。

flume+kafka+spark-streaming，应该说这一套架构已经成为流式计算的标配了。

如何整合我这里就不赘述了。

上几个配置文件好了

a1.sources = r1
a1.channels = c1
a1.sinks=k1

a1.sources.r1.type = TAILDIR
a1.sources.r1.channels = c1
a1.sources.r1.positionFile = /work/onedesk/bidlog/apache-flume-1.7.0-bin/taildir_position.json
a1.sources.r1.filegroups = f1 f2 f3
a1.sources.r1.filegroups.f1 = /work/onedesk/bidlog/bid.tmp
a1.sources.r1.headers.f1.topic = bid
a1.sources.r1.filegroups.f2 = /work/onedesk/bidlog/sspbid.tmp
a1.sources.r1.headers.f2.topic = sspbid
a1.sources.r1.filegroups.f3 = /work/onedesk/bidlog/sspclick.tmp
a1.sources.r1.headers.f3.topic = sspclick
a1.sources.r1.fileHeader = true


a1.channels.c1.type = memory
a1.channels.c1.capacity = 100000
a1.channels.c1.transactionCapacity = 1000

a1.sinks.k1.type = org.apache.flume.sink.kafka.KafkaSink
a1.sinks.k1.kafka.topic = bidlog
a1.sinks.k1.kafka.bootstrap.servers = 192.168.60.15:9092
a1.sinks.k1.kafka.flumeBatchSize = 20
a1.sinks.k1.kafka.producer.acks = 1
a1.sinks.k1.kafka.producer.linger.ms = 1
a1.sinks.k1.useFlumeEventFormat = true
a1.sinks.k1.channel=c1

a1.sources.r1.interceptors = i1 i2
a1.sources.r1.interceptors.i1.type = host
a1.sources.r1.interceptors.i1.hostHeader = hostname
a1.sources.r1.interceptors.i2.type = timestamp

val kafkaParams = Map[String, Object](
        "bootstrap.servers" -> "192.168.60.15:9092",
        "key.deserializer" -> classOf[StringDeserializer],
        "value.deserializer" -> classOf[BytesDeserializer],
        "group.id" -> "use_a_separate_group_id_for_each_stream",
        "auto.offset.reset" -> "latest",
        "enable.auto.commit" -> (false: java.lang.Boolean))
      val topics = Array("sspclick")
      val sspclick = KafkaUtils.createDirectStream[String, Bytes](
        ssc,
        PreferConsistent,
        Subscribe[String, Bytes](topics, kafkaParams))

总体的流程就是 flume会tail文件》将tail到的内容输出到Kafka中》spark订阅topic

默认模式下，Spark-streaming只能拿到flume tail到的文字，但是某些情况下我们希望spark也能处理一些header中的内容。

譬如说上述配置文件中在tail的同时，我们还像header中添加了timestamp以及host信息。

核心代码有两段：

第一点就是flume配置文件中

a1.sinks.k1.useFlumeEventFormat = true

该配置文件使得flume将整个信息封装成一个flumeEvent发送给kafka,这个flumeEvent就包含了header信息，默认是不包含的.

第二点就是spark在接收数据的时候

        "value.deserializer" -> classOf[BytesDeserializer],

      val sspclick = KafkaUtils.createDirectStream[String, Bytes](

接收的是一个byte数组，默认是string类型。

第三点就是讲byte数组转化成flumeEvent.

import org.apache.kafka.common.serialization.StringDeserializer
import org.apache.spark.streaming.kafka010._
import org.apache.kafka.common.serialization.BytesDeserializer
import org.apache.avro.specific.SpecificDatumReader
import org.apache.flume.source.avro.AvroFlumeEvent
import org.apache.avro.io.DecoderFactory
import org.apache.kafka.common.utils.Bytes
import org.apache.commons.lang3.StringUtils
import org.apache.avro.util.Utf8
import java.text.SimpleDateFormat
import java.sql.Timestamp
import java.sql.Date
import org.apache.avro.util.Utf8
import org.apache.kafka.common.serialization.BytesDeserializer
import org.apache.kafka.common.serialization.StringDeserializer
import org.apache.spark.streaming.dstream.DStream.toPairDStreamFunctions
import scala.reflect.runtime.universe
import org.apache.spark.sql.SparkSession
import org.apache.spark.sql.SaveMode
import org.apache.spark.streaming.kafka010._
import org.apache.spark.streaming.kafka010.LocationStrategies.PreferConsistent
import org.apache.spark.streaming.kafka010.ConsumerStrategies.Subscribe

  val reader = new SpecificDatumReader[AvroFlumeEvent](classOf[AvroFlumeEvent])

  val a = sspclick.map(f => {
        var body = f.value().get
        var decoder = DecoderFactory.get().binaryDecoder(body, null);
        var result = reader.read(null, decoder);
        var hostname = result.getHeaders.get(new Utf8("hostname"))
        var text = new String(result.getBody.array())
        var array = text.split("\t")

text就是tail中的内容。