You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@tajo.apache.org by "Hyunsik Choi (JIRA)" <ji...@apache.org> on 2013/06/12 09:49:19 UTC

[jira] [Updated] (TAJO-51) Parallel Container Launch of TaskRunnerLauncherImpl

     [ https://issues.apache.org/jira/browse/TAJO-51?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Hyunsik Choi updated TAJO-51:
-----------------------------

    Attachment: TAJO-51.patch

This patch does as follows:
 * Rename Container to ContainerProxy
 * Add a config AM_TASKRUNNER_LAUNCH_PARALLEL_NUM to TajoConf
 * Rename TaskRunnerEvent to TaskRunnerGroupEvent
 * Change TaskRunnerLauncherImpl to handle a group event for launching or killing containers
 * Move some methods of TaskRunnerLauncherImpl to ContainerProxy

I've also submitted the patch to review board.
https://reviews.apache.org/r/11827/
                
> Parallel Container Launch of TaskRunnerLauncherImpl
> ---------------------------------------------------
>
>                 Key: TAJO-51
>                 URL: https://issues.apache.org/jira/browse/TAJO-51
>             Project: Tajo
>          Issue Type: Improvement
>          Components: master
>            Reporter: Hyunsik Choi
>            Assignee: Hyunsik Choi
>              Labels: yarn
>         Attachments: TAJO-51.patch
>
>
> TaskRunnerLauncherImpl plays a role to launch remote containers via the proxy of ContainerManager. This implementation sequentially launches containers, so it may take a long time in large clusters. For example, it consumes about 5 seconds for starting 128 containers in our cluster, consisting of 32 nodes. 
> To relieve this time-consuming work, we can make this part parallel by using ExecutorService.

--
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