-Re: Preferred ways to specify input and output directories to Hadoop jobs
From what I know, a lot of guys including us provide the input and output dirs as position arguments for our custom mapreduce jobs along with other position arguments required for each job. These position arguments are assigned to mapred.input.dir, mapred.output.dir and other mapred params in our custom driver classes for each jobs.
Bejoy K S
From handheld, Please excuse typos.
From: "W.P. McNeill" <[EMAIL PROTECTED]>
Date: Wed, 8 Feb 2012 10:17:53
To: <[EMAIL PROTECTED]>; <[EMAIL PROTECTED]>
Subject: Re: Preferred ways to specify input and output directories to Hadoop jobs
Right. There is no default mechanism in Hadoop for using positional
arguments as input/output directory parameters. What I'm wondering is if
other people have done like me and written this mechanism themselves.