[
https://issues.apache.org/jira/browse/HIVE-1617?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Paul Yang updated HIVE-1617:
----------------------------
Attachment: HIVE-1617.1.patch
This patch can be tested with the following commands:
set hive.auto.progress.timeout=1200;
set hive.exec.script.allow.partial.consumption=true;
set hive.exec.mode.local.auto=false;
set hive.script.auto.progress=true;
select transform(*) using 'sleep 3600' from src;
The logs should show reporter messages at regular intervals, up to the 20
minute mark. Then the task should time out.
> ScriptOperator's AutoProgressor can lead to an infinite loop
> ------------------------------------------------------------
>
> Key: HIVE-1617
> URL: https://issues.apache.org/jira/browse/HIVE-1617
> Project: Hadoop Hive
> Issue Type: Improvement
> Components: Query Processor
> Reporter: Namit Jain
> Assignee: Paul Yang
> Fix For: 0.7.0
>
> Attachments: HIVE-1617.1.patch
>
>
> In the default settings, the auto progressor can result in a infinite loop.
> There should be another configurable parameter which stops the auto progress
> if the script has not made an progress.
> The default can be an hour or so - this way we will not get indefinitely stuck
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
