[jira] Commented: (HADOOP-4549) Cluster summary should have total tasks in the JT and pending tasks to run.

["Vinod K V (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-4549?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12643955#action_12643955 ] 

Vinod K V commented on HADOOP-4549:
-----------------------------------

Capacity scheduler maintains this information within its own map and reduce schedulers. But, with HADOOP-4445, they are being removed from the scheduler information as there are discrepancies between the running tasks' count it maintains and the one in the cluster-summary. These two counts represent the same information and currently there is no way to synchronize these two counts all the time. The place from where we can obtain consistent information is in JT's cluster-summary.

> Cluster summary should have total tasks in the JT and pending tasks to run.
> ---------------------------------------------------------------------------
>
>                 Key: HADOOP-4549
>                 URL: https://issues.apache.org/jira/browse/HADOOP-4549
>             Project: Hadoop Core
>          Issue Type: Improvement
>          Components: mapred
>            Reporter: Vinod K V
>            Assignee: Devaraj Das
>             Fix For: 0.20.0
>
>
> Cluster summary already shows the running maps and running reduces. It is useful to have total tasks from all initialized jobs and pending tasks from initialized jobs that still need to be run. Pending tasks can be derived from total tasks and running tasks.

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