[jira] [Commented] (SPARK-20935) A daemon thread, "BatchedWriteAheadLog Writer", left behind after terminating StreamingContext.

["Hyukjin Kwon (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SPARK-20935?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16034035#comment-16034035 ] 

Hyukjin Kwon commented on SPARK-20935:
--------------------------------------

Thanks for pinging me. Could we just always stop() {{ReceivedBlockTracker}} and require {{WriteAheadLog.close()}} to be idempotent and make its implementations as so? I could propose a PR if it sounds okay instead.

> A daemon thread, "BatchedWriteAheadLog Writer", left behind after terminating StreamingContext.
> -----------------------------------------------------------------------------------------------
>
>                 Key: SPARK-20935
>                 URL: https://issues.apache.org/jira/browse/SPARK-20935
>             Project: Spark
>          Issue Type: Bug
>          Components: DStreams
>    Affects Versions: 1.6.3, 2.1.1
>            Reporter: Terence Yim
>
> With batched write ahead log on by default in driver (SPARK-11731), if there is no receiver based {{InputDStream}}, the "BatchedWriteAheadLog Writer" thread created by {{BatchedWriteAheadLog}} never get shutdown. 
> The root cause is due to https://github.com/apache/spark/blob/master/streaming/src/main/scala/org/apache/spark/streaming/scheduler/ReceiverTracker.scala#L168
> that it never call {{ReceivedBlockTracker.stop()}} (which in turn call {{BatchedWriteAheadLog.close()}}) if there is no receiver based input.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

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