[
https://issues.apache.org/jira/browse/HADOOP-18221?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17532233#comment-17532233
]
Steve Loughran commented on HADOOP-18221:
-----------------------------------------
if the draining can be done asynchronously, this would benefit worker threads
which want to finish their work and report completion to their driver processes
and/or start on the next input file.
But: we would have to make sure that the block being read wasn't persisted, as
it will be incomplete
> stream warns Not all bytes were read from the S3ObjectInputStream when closed
> -----------------------------------------------------------------------------
>
> Key: HADOOP-18221
> URL: https://issues.apache.org/jira/browse/HADOOP-18221
> Project: Hadoop Common
> Issue Type: Sub-task
> Reporter: Ahmar Suhail
> Assignee: Ahmar Suhail
> Priority: Minor
>
> Issue: [https://github.com/aws/aws-sdk-java/issues/1211] has resurfaced in
> the prefetching stream when it is closed before reading for blocks is
> complete. This can be fixed by draining the stream before closingÂ
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
