[
https://issues.apache.org/jira/browse/HIVE-4090?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kevin Wilfong updated HIVE-4090:
--------------------------------
Attachment: HIVE-4090.1.patch.txt
> Use of hive.exec.script.allow.partial.consumption can produce partial results
> -----------------------------------------------------------------------------
>
> Key: HIVE-4090
> URL: https://issues.apache.org/jira/browse/HIVE-4090
> Project: Hive
> Issue Type: Bug
> Components: Query Processor
> Affects Versions: 0.11.0
> Reporter: Kevin Wilfong
> Attachments: HIVE-4090.1.patch.txt
>
>
> When users execute use a transform script with the config
> hive.exec.script.allow.partial.consumption set to true, it may produce
> partial results.
> When this config is set the script may close it's input pipe before its
> parent operator has finished passing it rows. In the catch block for this
> exception, the setDone method is called marking the operator as done.
> However, there's a separate thread running to process rows passed from the
> script back to Hive via stdout. If this thread is not done processing rows,
> any rows it forwards after the setDone method is called will not be passed to
> its children. This leads to partial results.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
