推送式
- 将flume采集的数据主动推送给Spark程序，容易导致Spark程序接受数据出问题，推送式整合是基于avro端口下沉地方式完成
- 引入SparkStreaming和Flume整合的依赖
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-streaming_2.11</artifactId>
<version>2.3.1</version>
</dependency>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-streaming-flume_2.11</artifactId>
<version>2.3.1</version>
</dependency>
</dependencies>
- 定义Flume采集数据进程脚本，把sink下沉地指定为avro类型的端口下沉底
[root@node1 data]# vi portToSpark.conf
#指定agent的sources,sinks,channels
a1.sources = s1
a1.sinks = k1
a1.channels = c1
#配置sources属性
a1.sources.s1.type = netcat
a1.sources.s1.bind = node1
a1.sources.s1.port = 44444
#配置sink
a1.sinks.k1.type = avro
a1.sinks.k1.hostname = node1
a1.sinks.k1.port = 8888
a1.sinks.k1.batch-size = 1
#配置channel类型
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 100
#整合flume进程中source channel sink
a1.sources.s1.channels = c1
a1.sinks.k1.channel = c1
- 通过FileUtils.createStream方法从avro的端口中获取flume采集到avro端口的实时数据
package SparkStreaming.flume
import org.apache.spark.SparkConf
import org.apache.spark.storage.StorageLevel
import org.apache.spark.streaming.flume.FlumeUtils
import org.apache.spark.streaming.{Seconds, StreamingContext}
object ByFlumePush {
def main(args: Array[String]): Unit = {
val conf = new SparkConf().setMaster("local[3]").setAppName("hdfs")
val ssc: StreamingContext = new StreamingContext(conf, Seconds(10))
val ds = FlumeUtils.createStream(ssc, "node1", 8888, StorageLevel.MEMORY_ONLY)
ds.print()
ssc.start()
ssc.awaitTermination()
}
}
- 启动
1. 启动flume
flume-ng agent -n a1 -f portToSpark.conf -Dflume.root.logger=INFO
2. 运行主类,将java代码打包上传到node1上
spark-submit --class flume.Demo01 ssc.jar
3. 开启监听的端口号
[root@node1 ~]# telnet node1 44444
- 注意：
必须保证Spark Streaming运行程序和Flume采集进程在同一个节点上，保证Spark Streaming打包的jar包必须把spark-streaming-flume_2.11:2.3.1版本的依赖包全部打包到jar包中
(这里用的别人打的包，保存在G://shixun//ssc.jar路径下了)
拉取式
- 将Flume采集的数据发送给sink了，sink并不是直接把数据立马给了Spark，而是先把数据缓冲，Spark接收器可以按照我的需求主动去sink中拉取数据.
拉取式整合方式是基于Spark下沉地完成----建议使用 - 引入依赖：
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-streaming_2.11</artifactId>
<version>2.3.1</version>
</dependency>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-streaming-flume_2.11</artifactId>
<version>2.3.1</version>
</dependency>
</dependencies>
- 定义flume脚本文件，和上面的方式同，但把sink的下沉地改为SparkSink
[root@node1 data]# vi portToSpark2.conf
#指定agent的sources,sinks,channels
a1.sources = s1
a1.sinks = k1
a1.channels = c1
#配置sources属性
a1.sources.s1.type = netcat
a1.sources.s1.bind = node1
a1.sources.s1.port = 44444
#配置sink
a1.sinks.k1.type = org.apache.spark.streaming.flume.sink.SparkSink
a1.sinks.k1.hostname = node1
a1.sinks.k1.port = 8888
a1.sinks.k1.batch-size = 1
#配置channel类型
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 100
#整合flume进程中source channel sink
a1.sources.s1.channels = c1
a1.sinks.k1.channel = c1
- 定义读取方法
package SparkStreaming.flume
import org.apache.spark.SparkConf
import org.apache.spark.storage.StorageLevel
import org.apache.spark.streaming.flume.FlumeUtils
import org.apache.spark.streaming.{Seconds, StreamingContext}
object ByFlumePush {
def main(args: Array[String]): Unit = {
val conf = new SparkConf().setMaster("local[3]").setAppName("hdfs")
val ssc: StreamingContext = new StreamingContext(conf, Seconds(10))
val ds = FlumeUtils.createPollingStream(ssc, "node1", 8888, StorageLevel.MEMORY_ONLY)
ds.print()
ssc.start()
ssc.awaitTermination()
}
}
- [注意]：
SparkStreaming的依赖jar包复制到flume软件的lib目录下，把spark-streaming-flume的依赖jar包放到flume软件的lib目录下
[root@node1 jars]# pwd
/opt/app/spark-2.3.1/jars
[root@node1 jars]# cp spark-streaming_2.11-2.3.1.jar /opt/app/flume-1.8.0/lib/
[root@node1 data]# pwd
/opt/data
[root@node1 data]# cp ssc.jar /opt/app/flume-1.8.0/lib/
(ssc.jar为别人打的包，保存在G://shixun//ssc.jar路径下了)
- 启动
[root@node1 data]# flume-ng agent -n a1 -f portToSpark2.conf -Dflume.root.logger=INFO,console
[root@node1 data]# spark-submit --class flume.ByFlumePush ssc2.jar
[root@node1 data]# telnet node1 44444