You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Kristam Subba Swathi (Created) (JIRA)" <ji...@apache.org> on 2012/04/11 07:36:41 UTC
[jira] [Created] (HIVE-2945) Hive Client is hanging since MRApp
Master doesnt exist after the task is finished
Hive Client is hanging since MRApp Master doesnt exist after the task is finished
----------------------------------------------------------------------------------
Key: HIVE-2945
URL: https://issues.apache.org/jira/browse/HIVE-2945
Project: Hive
Issue Type: Bug
Components: Clients
Reporter: Kristam Subba Swathi
Submit a hive query which uses mapred to give the result(Say join query)
If we submit a job to resourceManager process. it will create a MRAppMaster process to excute this task .
MRAppMaster will create YarnChild process to excute Map task or reduce task.
after task is finished,resourceManager will kill MRAppMaster and YarnChild process.
When the hive client is about to get the progress from MRAppMaster ,the task is already finished ,so MRAppMaster process don't exist. so the client will try to connect it for ever.
that's why hive client hangs.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HIVE-2945) Hive Client is hanging since MRApp
Master doesnt exist after the task is finished
Posted by "Edward Capriolo (Commented) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HIVE-2945?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13251616#comment-13251616 ]
Edward Capriolo commented on HIVE-2945:
---------------------------------------
I am not sure anyone has tried hive in the Mapred.NEXT setup yet. I do not know enough about the infrastructure to comment definitively , but it sounds like it would be nice if the processes hung around for a few moments after completion so Hive client could check in to them. Strange though because you would not expect a hive job to hang around long then a typical map/reduce job.
It would be great if you can give a stack trace and find what point in the hive code is hanging.
> Hive Client is hanging since MRApp Master doesnt exist after the task is finished
> ----------------------------------------------------------------------------------
>
> Key: HIVE-2945
> URL: https://issues.apache.org/jira/browse/HIVE-2945
> Project: Hive
> Issue Type: Bug
> Components: Clients
> Reporter: Kristam Subba Swathi
>
> Submit a hive query which uses mapred to give the result(Say join query)
> If we submit a job to resourceManager process. it will create a MRAppMaster process to excute this task .
> MRAppMaster will create YarnChild process to excute Map task or reduce task.
> after task is finished,resourceManager will kill MRAppMaster and YarnChild process.
> When the hive client is about to get the progress from MRAppMaster ,the task is already finished ,so MRAppMaster process don't exist. so the client will try to connect it for ever.
> that's why hive client hangs.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira