You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by "Hemanth Yamijala (JIRA)" <ji...@apache.org> on 2008/07/23 08:28:31 UTC

[jira] Commented: (HADOOP-3808) [HOD] Include job tracker RPC in notes attribute after job submission

    [ https://issues.apache.org/jira/browse/HADOOP-3808?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12615893#action_12615893 ] 

Hemanth Yamijala commented on HADOOP-3808:
------------------------------------------

The attached patch makes the following changes:

- Modified signature of updateWorkerInfo in nodePool.py as follows - including the jobId, which should already have been there, and a list of keys - which decides the order in which they are written to the notes attribute:
{code}
def updateWorkerInfo(self, workerInfoMap, jobId, keys)
{code}

- Modified implementation of the updateWorkerInfo in torque.py to use the new signature.

- In hadoop.py, calling the method with the RPC port.

 

> [HOD] Include job tracker RPC in notes attribute after job submission
> ---------------------------------------------------------------------
>
>                 Key: HADOOP-3808
>                 URL: https://issues.apache.org/jira/browse/HADOOP-3808
>             Project: Hadoop Core
>          Issue Type: Improvement
>    Affects Versions: 0.18.0
>            Reporter: Hemanth Yamijala
>            Assignee: Hemanth Yamijala
>         Attachments: HADOOP-3808.patch
>
>
> Currently, after a job is submitted, HOD records the Jobtracker URL and the NameNode URL in the 'notes' attribute of the Torque resource manager. This helps in building centralized administration tools that need to get a full picture of all allocated clusters etc. Including the Jobtracker RPC port in this attribute would be useful to build such similar tools. One of them is an idle job tracker that can be centrally run to clear up any clusters that aren't automatically cleaned up due to some faulty nodes in the cluster or resource manager problems.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.