[
https://issues.apache.org/jira/browse/SPARK-16246?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15352470#comment-15352470
]
Sean Owen commented on SPARK-16246:
-----------------------------------
You'd have to say a lot more about how you're running this, including the
number of partitions, workers, and where the thread is waiting (stack trace)
> Too many block-manager-slave-async-thread opened (TIMED_WAITING) for spark
> Kafka streaming
> ------------------------------------------------------------------------------------------
>
> Key: SPARK-16246
> URL: https://issues.apache.org/jira/browse/SPARK-16246
> Project: Spark
> Issue Type: Bug
> Components: Spark Core, Streaming
> Affects Versions: 1.6.1
> Reporter: Alex Jiang
>
> I don't know if our spark streaming issue is related to this
> (https://issues.apache.org/jira/browse/SPARK-15558).
> Basically we have one Kafka receiver on each executor, and it ran fine for a
> while. Then, the executor had a lot of waiting thread accumulated (Thread
> 1224: block-manager-slave-async-thread-pool-1083 (TIMED_WAITING)). And the
> executor kept open such new thread. Eventually, it reached the maximum number
> of the thread on that executor and Kafka receiver on that executor failed.
> Could someone please shed some light on this?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
