hadoop-mapreduce-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Roelofs <roel...@yahoo-inc.com>
Subject Re: Changing default separator for streaming application
Date Thu, 17 Jun 2010 20:07:26 GMT
Ravi Chinni wrote:

> When invoking the application I specified  -D
> stream.map.output.field.separator=" " -D
> stream.reduce.output.field.seperator=" " options.

> While in the output of my shell script I have a space between key and
> value fields, the final output written by the framework to file has a
> tab as the separator. It seems that the framework is replacing the space
> separator by a tab separator in the output of mapper and reducer
> functions.

Both of them, or just the reducer half?  A mismatch between the mapper's
output and the reducer's expected input format presumably would result in
completely bogus results, wouldn't it?  Or is the value of stream.map.output
.field.separator implicitly also the value of the reducer's input separator?
(Sorry, I'm still new to this stuff.)

> If anyone has ideas on how I can fix this, please share it.

Spell "separator" correctly? :-)


View raw message