Overview
- Flume：一个分布式的，可靠的，可用的服务，用于有效地收集、聚合、移动大规模日志数据
- 我们搭建一个flume + Spark Streaming的平台来从Flume获取数据，并处理它。
- 有两种方法实现：使用flume-style的push-based方法，或者使用自定义的sink来实现pull-based方法。
Approach 1: Flume-style Push-based Approach
- flume被设计用来在Flume agents之间推信息，在这种方式下，Spark Streaming安装一个receiver that acts like an Avro agent for Flume, to which Flume can push the data.
General Requirement
- 当你启动flume + spark streaming应用时，该机器上必须运行一个Spark workers。
- flume可以向该机器的某一个port push数据。
- 基于这种push机制，streaming应用必须有一个receiver scheduled and listening on the chosen port.
Configuring Flume
- 配置flume以向Avro sink发送数据
-
agent.sinks = avroSinkagent.sinks.avroSink.type = avroagent.sinks.avroSink.channel = memoryChannelagent.sinks.avroSink.hostname =
agent.sinks.avroSink.port =
Configuring Spark Streaming Application
- Linking: 在maven项目中配置依赖
org.apache.spark spark-streaming-flume-sink_2.10 2.1.0
2. Programming：import FlumeUtils, 创建input DStream
import org.apache.spark.streaming.flume._ val flumeStream = FlumeUtils.createStream(streamingContext, [chosen machine's hostname], [chosen port])
- 注意：应该与cluster中的resourceManager使用同一个hostname，这样的话资源分配可以匹配names，并在正确的机器上launch receiver
- 一个简单的Spark Streaming统计Flume event个数的demo代码：
-
object FlumeEventCount { def main(args: Array[String]) { if (args.length < 2) { System.err.println( "Usage: FlumeEventCount") System.exit(1) } StreamingExamples.setStreamingLogLevels() val Array(host, IntParam(port)) = args val batchInterval = Milliseconds(2000) // Create the context and set the batch size val sparkConf = new SparkConf().setAppName("FlumeEventCount") val ssc = new StreamingContext(sparkConf, batchInterval) // Create a flume stream val stream = FlumeUtils.createStream(ssc, host, port, StorageLevel.MEMORY_ONLY_SER_2) // Print out the count of events received from this server in each batch stream.count().map(cnt => "Received " + cnt + " flume events." ).print() ssc.start() ssc.awaitTermination() }}