[jira] [Updated] (SPARK-32735) RDD actions in DStream.transfrom don't show at batch page

["Takeshi Yamamuro (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/SPARK-32735?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Takeshi Yamamuro updated SPARK-32735:
-------------------------------------
    Labels:   (was: pull-request-available)

> RDD actions in DStream.transfrom don't show at batch page
> ---------------------------------------------------------
>
>                 Key: SPARK-32735
>                 URL: https://issues.apache.org/jira/browse/SPARK-32735
>             Project: Spark
>          Issue Type: Bug
>          Components: DStreams, Web UI
>    Affects Versions: 3.0.0
>            Reporter: Liechuan Ou
>            Priority: Major
>
> h4. Issue
> {code:java}
> val lines = ssc.socketTextStream("localhost", 9999)
> val words = lines.flatMap(_.split(" "))
> val mappedStream= words.transform(rdd => {
>   val c = rdd.count();
>   rdd.map(x => s"$c x")}
> )
> mappedStream.foreachRDD(rdd => rdd.foreach(x => println(x))){code}
> Every batch two spark jobs are created. Only the second one is associated with the streaming output operation and shows at batch page.
> h4. Investigation
> The first action rdd.count() is invoked by JobGenerator.generateJobs. Batch time and output op id are not available in spark context because they are set in JobScheduler later.
> h4. Proposal
> delegate dstream.getOrCompute to JobScheduler so that all rdd actions can run in spark context with correct local properties.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org