You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hama.apache.org by "Suraj Menon (Updated) (JIRA)" <ji...@apache.org> on 2012/03/17 20:11:37 UTC

[jira] [Updated] (HAMA-533) BSP Peer should have the ability to start with a non-zero superstep from a partition of checkpointed message for that task ID, attempt ID

     [ https://issues.apache.org/jira/browse/HAMA-533?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Suraj Menon updated HAMA-533:
-----------------------------

    Issue Type: Sub-task  (was: New Feature)
        Parent: HAMA-505
    
> BSP Peer should have the ability to start with a non-zero superstep from a partition of checkpointed message for that task ID, attempt ID
> -----------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: HAMA-533
>                 URL: https://issues.apache.org/jira/browse/HAMA-533
>             Project: Hama
>          Issue Type: Sub-task
>          Components: bsp
>    Affects Versions: 0.5.0
>            Reporter: Suraj Menon
>             Fix For: 0.5.0
>
>
> Currently BSP Peer initializes itself to start a task afresh from superstep -1. We should have a new flavor of BSP Peer that would be started by GroomServer to recover a task as directed by BSP Master. The BSP Peer should start the task knowing the superstep number to start with, the task ID, the attemptID and the partition of the checkpointed file. The input would not be split in this case. The BSP Peer should update its task status at Groomserver as RUNNING once the task is out of the recovery superstep barrier sync.

--
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