SparkStreaming 集成Kafka详解
1.概述
SparkStream集成Kafa有两种方式,Receiver-based Approach与Direct Approach,在集成的时候得注意spark版本与kafka的版本是否兼容。
2.Receiver-based Approach
1)概述
此种方式使用的是recevier 接收数据,recevier使用的是the Kafka high-level consumer API,默认配置可能会丢失部分数据,使用Write Ahead Logs把日志文件写到文件系统中(hdfs),数据丢失可以从日志文件中恢复。
2)引入jar包
groupId = org.apache.spark
artifactId = spark-streaming-kafka-0-8_2.11
version = 2.1.1
3)编码
def main(args: Array[String]): Unit = {
val conf = new SparkConf().setMaster("local[2]").setAppName("SparkStreamingBykafkaRecevier")
val ssc = new StreamingContext(conf, Seconds(5))
var topicsMap = Map[String, Int]("test20" -> 1)
val dstream = KafkaUtils.createStream(ssc, "192.168.126.31:2181,192.168.126.32:2181,192.168.126.33:2181", "my_group_reciver", topicsMap)
val lines = dstream.map(_._2)
val wordcount = lines.flatMap(_.split(" ")).map(x=>(x,1)).reduceByKey(_+_).print()
ssc.start()
ssc.awaitTermination()
}
4)测试
启动zookeeper: zkServer.sh start
启动kafka: kafka-server-start.sh -daemon $KAFKA_HOME/conf/server.properties
测试kafka集群是否可用
创建topic:
kafka-topics --create --zookeeper master:2181,slave1:2181,slave2:2181 --replication-factor 1 --partitions 1 --topics test20
查看topic:
kafka-topics --list --zookeeper master:2181,slave1:2181,slave2:2181
生产者生成消息:
kafka-console-producer.sh --broker-list master:9092 --topic test20
消费者消费消息:
kafka-console-consumer.sh --zookeeper master:2181,slave2:2181 --topics test20 --from begginning
本地运行项目:在集群生成者上输入数据,看运行控制台是否接收到数据
5)正式环境
./bin/spark-submit \
--class com.zhm.sparkstreaming.kafka.SparkStreamingBykafkaRecevier \
--master local[2] \
--packages org.apache.spark:spark-streaming-kafka-0-8_2.11:2.1.1 \
jar路径
3.Direct Approach (No Receivers)
1)概述
receiver-less “direct” approach 在spark1.3引入的,此种方式周期性的到kafka获取最新偏移量(offsets),Kafka’s simple consumer API 从偏移量中获取批次。
2)导入包
groupId = org.apache.spark
artifactId = spark-streaming-kafka-0-8_2.11
version = 2.1.1
3)编码
def main(args: Array[String]): Unit = {
val conf = new SparkConf().setMaster("local[2]").setAppName("SparkStreamByKafkaDirect")
val ssc = new StreamingContext(conf,Seconds(5))
val kafkaParams = Map[String,String]("metadata.broker.list"-> "192.168.126.31:9092")
val topics = Set("test20")
val dstream = KafkaUtils.createDirectStream[String,String,StringDecoder,StringDecoder](ssc,kafkaParams,topics)
val lines = dstream.map(_._2)
lines.flatMap(_.split(" ")).map((_,1)).reduceByKey(_+_).print()
ssc.start()
ssc.awaitTermination()
}
4)测试
启动zookeeper: zkServer.sh start
启动kafka: kafka-server-start.sh -daemon $KAFKA_HOME/conf/server.properties
测试kafka集群是否可用
创建topic:
kafka-topics --create --zookeeper master:2181,slave1:2181,slave2:2181 --replication-factor 1 --partitions 1 --topics test20
查看topic:
kafka-topics --list --zookeeper master:2181,slave1:2181,slave2:2181
生产者生成消息:
kafka-console-producer.sh --broker-list master:9092 --topic test20
消费者消费消息:
kafka-console-consumer.sh --zookeeper master:2181,slave2:2181 --topics test20 --from begginning
本地运行项目:在集群生成者上输入数据,看运行控制台是否接收到数据
正式环境:
./bin/spark-submit \
--class com.zhm.sparkstreaming.kafka.SparkStreamByKafkaDirect \
--master local[2] \
--packages org.apache.spark:spark-streaming-kafka-0-8_2.11:2.1.1 \
jar路径
注意: 在测试环境中使用–packages 方便,在生产环境中一般是没网的,需要下载jar包放到spark的libs中
spark-streaming-kafka-0-8-assembly jar包下载地址:https://search.maven.org/search?q=a:spark-streaming-kafka-0-8-assembly_2.11 AND v:2.1.1
在使用spark-submit 提交的时候使用: --jars 导入jar包