Question

我有一个火花流式传输作业，每隔5秒从Kafka读取一次，对传入数据进行一些转换，然后写入文件系统。

这实际上并不需要是一个流媒体工作，实际上，我只想每天运行一次以将消息排入文件系统。我不知道如何阻止这份工作。

如果我将超时传递给streamingContext.awaitTermination，它不会停止进程，它只会导致进程在流上进行迭代时产生错误（请参阅下面的错误）

完成我想要做的事情的最佳方法是什么

这适用于Python上的Spark 1.6

编辑：

感谢@marios解决方案就是：

ssc.start()
ssc.awaitTermination(10)
ssc.stop()

在停止之前运行脚本十秒钟。

简化代码：

conf = SparkConf().setAppName("Vehicle Data Consolidator").set('spark.files.overwrite','true')
sc = SparkContext(conf=conf)
ssc = StreamingContext(sc, 5)
stream = KafkaUtils.createStream(
    ssc,
    kafkaParams["zookeeper.connect"],
    "vehicle-data-importer",
    topicPartitions,
    kafkaParams)

stream.saveAsTextFiles('stream-output/kafka-vehicle-data')

ssc.start()
ssc.awaitTermination(10)

错误：

16/01/29 15:05:44 INFO BlockManagerInfo: Added input-0-1454097944200 in memory on localhost:58960 (size: 3.0 MB, free: 48.1 MB)
16/01/29 15:05:44 WARN BlockManager: Block input-0-1454097944200 replicated to only 0 peer(s) instead of 1 peers
16/01/29 15:05:44 INFO BlockGenerator: Pushed block input-0-1454097944200
16/01/29 15:05:45 ERROR JobScheduler: Error generating jobs for time 1454097945000 ms
py4j.Py4JException: Cannot obtain a new communication channel
    at py4j.CallbackClient.sendCommand(CallbackClient.java:232)
    at py4j.reflection.PythonProxyHandler.invoke(PythonProxyHandler.java:111)
    at com.sun.proxy.$Proxy14.call(Unknown Source)
    at org.apache.spark.streaming.api.python.TransformFunction.callPythonTransformFunction(PythonDStream.scala:92)
    at org.apache.spark.streaming.api.python.TransformFunction.apply(PythonDStream.scala:78)
    at org.apache.spark.streaming.api.python.PythonTransformedDStream.compute(PythonDStream.scala:230)
    at org.apache.spark.streaming.dstream.DStream$$anonfun$getOrCompute$1$$anonfun$1$$anonfun$apply$7.apply(DStream.scala:352)
    at org.apache.spark.streaming.dstream.DStream$$anonfun$getOrCompute$1$$anonfun$1$$anonfun$apply$7.apply(DStream.scala:352)
    at scala.util.DynamicVariable.withValue(DynamicVariable.scala:57)
    at org.apache.spark.streaming.dstream.DStream$$anonfun$getOrCompute$1$$anonfun$1.apply(DStream.scala:351)
    at org.apache.spark.streaming.dstream.DStream$$anonfun$getOrCompute$1$$anonfun$1.apply(DStream.scala:351)
    at org.apache.spark.streaming.dstream.DStream.createRDDWithLocalProperties(DStream.scala:426)
    at org.apache.spark.streaming.dstream.DStream$$anonfun$getOrCompute$1.apply(DStream.scala:346)
    at org.apache.spark.streaming.dstream.DStream$$anonfun$getOrCompute$1.apply(DStream.scala:344)
    at scala.Option.orElse(Option.scala:257)
    at org.apache.spark.streaming.dstream.DStream.getOrCompute(DStream.scala:341)
    at org.apache.spark.streaming.dstream.ForEachDStream.generateJob(ForEachDStream.scala:47)
    at org.apache.spark.streaming.DStreamGraph$$anonfun$1.apply(DStreamGraph.scala:115)
    at org.apache.spark.streaming.DStreamGraph$$anonfun$1.apply(DStreamGraph.scala:114)
    at scala.collection.TraversableLike$$anonfun$flatMap$1.apply(TraversableLike.scala:251)
    at scala.collection.TraversableLike$$anonfun$flatMap$1.apply(TraversableLike.scala:251)
    at scala.collection.mutable.ResizableArray$class.foreach(ResizableArray.scala:59)
    at scala.collection.mutable.ArrayBuffer.foreach(ArrayBuffer.scala:47)
    at scala.collection.TraversableLike$class.flatMap(TraversableLike.scala:251)
    at scala.collection.AbstractTraversable.flatMap(Traversable.scala:105)
    at org.apache.spark.streaming.DStreamGraph.generateJobs(DStreamGraph.scala:114)
    at org.apache.spark.streaming.scheduler.JobGenerator$$anonfun$3.apply(JobGenerator.scala:248)
    at org.apache.spark.streaming.scheduler.JobGenerator$$anonfun$3.apply(JobGenerator.scala:246)
    at scala.util.Try$.apply(Try.scala:161)
    at org.apache.spark.streaming.scheduler.JobGenerator.generateJobs(JobGenerator.scala:246)
    at org.apache.spark.streaming.scheduler.JobGenerator.org$apache$spark$streaming$scheduler$JobGenerator$$processEvent(JobGenerator.scala:181)
    at org.apache.spark.streaming.scheduler.JobGenerator$$anon$1.onReceive(JobGenerator.scala:87)
    at org.apache.spark.streaming.scheduler.JobGenerator$$anon$1.onReceive(JobGenerator.scala:86)
    at org.apache.spark.util.EventLoop$$anon$1.run(EventLoop.scala:48)
16/01/29 15:05:45 INFO MemoryStore: Block input-0-1454097944800 stored as bytes in memory (estimated size 3.0 MB, free 466.1 MB)
16/01/29 15:05:45 INFO BlockManagerInfo: Added input-0-1454097944800 in memory on localhost:58960 (size: 3.0 MB, free: 45.1 MB)

Answer 1

似乎正确的调用方法是awaitTerminationOrTimeout（自我，超时）。

我不确定它是否也会停止流式上下文。所以也许你可以在超时结束后立即调用ssc.stop（）。

ssc.start()
ssc.awaitTerminationOrTimeout(10)
ssc.stop()

注意：请查看here以查找类似问题。

Answer 2

尝试使用Kafka“consumer.timeout.ms”参数，它将优雅地结束KafkaReceiver。（来自kafka 0.8 configuration）

如果没有可用消息，则向消费者抛出超时异常在指定的时间间隔后消费

HDF = KafkaUtils.createStream(ssc, topics={strLoc : 1}, kafkaParams={"consumer.timeout.ms":"20000" }, zkQuorum='xxx:2181', groupId='xxx-consumer-group')

您无法在当前的流式传输执行中收到任何新的kafka消息，并且始终会获得空的RDD 并检查DSteam.foreachRDD（func）中空RDD的计数。如果您不断获得空的RDD，请终止流媒体exectuion。

Answer 3

这里的问题在与Dstream处理线程相同的线程中触发ssc.stop调用1.6会创建死锁，因为stop将等待轮询器线程完成创建死锁。sp从另一个线程调用stop

如何在数据源用尽时停止火花流式传输

3 个答案: