[jira] Updated: (MAPREDUCE-1158) running_maps is not decremented when the tasks of a job is killed/failed

["Sharad Agarwal (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/MAPREDUCE-1158?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sharad Agarwal updated MAPREDUCE-1158:
--------------------------------------

    Attachment: 1158.patch

Currently runningMaps/runningReduces metrics is captured in JobInProgress. It was discussed in offline discussion with Hemanth and Amareshwari that right place would be to do in JobTracker where slot metrics is captured. Attaching patch doing the same.

> running_maps is not decremented when the tasks of a job is killed/failed 
> -------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-1158
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-1158
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>          Components: jobtracker
>    Affects Versions: 0.20.1
>            Reporter: Ramya R
>             Fix For: 0.20.2
>
>         Attachments: 1158.patch
>
>
> running_maps counter in the metrics is not decremented when the tasks of a job is killed/failed. Below are the exact steps to reproduce the problem:
> * Initially running_maps=0
> * Submit a job with 5 maps. running_maps is set to 5
> * Kill 2 attempts of a map task
> * Fail 4 attempts of the same map task so that the job is finally marked killed.
> * Once the job is marked killed, running_maps is set to 3 and not 0. 

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