hadoop-mapreduce-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Paul Vasilev (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (MAPREDUCE-6053) New hadoop API don't allow dynamic key-based names for output files
Date Tue, 26 Aug 2014 11:23:57 GMT

     [ https://issues.apache.org/jira/browse/MAPREDUCE-6053?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Paul Vasilev updated MAPREDUCE-6053:
------------------------------------

    Description: 
MultipleTextOutputFormat class removed from new api. MultipleOutputs class forces developer
to set names of output files at job configuration time.
So with new api I can't create files with names based on keys (I don't know all keys. Therefore
I can't set output file names at job configuration time).
This is major disadvantage in comparison with old api and force developer to use it.

  was:
MultipleTextOutputFormat class removed from new api. MultipleOutputs class force developer
to set names of output files at job configuration time. So with new api I can't create files
with names based on keys (I don't know all keys. Therefore I can't set output file names at
job configuration time).
This is major disadvantage in comparison with old api and force developer to use it.


> New hadoop API don't allow dynamic key-based names for output files
> -------------------------------------------------------------------
>
>                 Key: MAPREDUCE-6053
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-6053
>             Project: Hadoop Map/Reduce
>          Issue Type: Improvement
>          Components: job submission
>            Reporter: Paul Vasilev
>
> MultipleTextOutputFormat class removed from new api. MultipleOutputs class forces developer
to set names of output files at job configuration time.
> So with new api I can't create files with names based on keys (I don't know all keys.
Therefore I can't set output file names at job configuration time).
> This is major disadvantage in comparison with old api and force developer to use it.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message