[
https://issues.apache.org/jira/browse/SPARK-3018?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14098027#comment-14098027
]
Michael Yannakopoulos commented on SPARK-3018:
----------------------------------------------
Hi Reynold,
I am new to Spark, having contributed mostly to simple tasks in MLLIB. I am
interested in working on this issue if no one already does.
Thanks,
Michael
> Release all BlockFetcherIterator upon task completion/failure
> -------------------------------------------------------------
>
> Key: SPARK-3018
> URL: https://issues.apache.org/jira/browse/SPARK-3018
> Project: Spark
> Issue Type: Sub-task
> Components: Shuffle, Spark Core
> Reporter: Reynold Xin
>
> BlockFetcherIterator retains ReferenceCountedBuffers returned by
> client.fetchBlocks. Those buffers are released when the iterators are
> traversed fully. In the case of task failures or completion without
> exhausting the iterator, this could lead to memory leak.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
