[jira] [Commented] (MAPREDUCE-3678) The Map tasks logs should have the value of input split it processed

["Harsh J (Commented) (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MAPREDUCE-3678?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13187044#comment-13187044 ] 

Harsh J commented on MAPREDUCE-3678:
------------------------------------

Task's own logs are the best place for this, not the daemons.

The reason it is tedious to do/maintain at the framework level is that not all InputSplits may be FileSplits, and formats that do use FileSplits may use them in different ways as well (CombineFileIF, for instance).

The InputSplit interface by itself is path-agnostic.
                
> The Map tasks logs should have the value of input split it processed
> --------------------------------------------------------------------
>
>                 Key: MAPREDUCE-3678
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-3678
>             Project: Hadoop Map/Reduce
>          Issue Type: New Feature
>          Components: nodemanager, tasktracker
>    Affects Versions: 0.20.203.0, 0.20.205.0, 1.0.0
>         Environment: Linux red hat.
>            Reporter: Bejoy KS
>
> It would be easier to debug some corner in tasks if we knew what was the input split processed by that task. Map reduce task tracker log should accommodate the same. Also in the jobdetails web UI, the split also should be displayed along with the Split Locations. 
> Sample as
> Input Split
> hdfs://myserver:9000/userdata/sampleapp/inputdir/file1.csv - <split no>/<offset from beginning of file>
> This would be much beneficial to nail down some data quality issues in large data volume processing.

--
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