I am doing something similar where I have a million input files to process, I provide a single text file as input and use NLineInputFormat, specifying the number of lines that are sent to each map task.
In your case you would set it so that 1 line was one input split, which is the default. In pre 20, the parameter is mapred.line.input.format.linespermap. On Wed, Oct 14, 2009 at 10:36 PM, Something Something < luckyguy2...@yahoo.com> wrote: > If the answer is... > > TableMapReduceUtil.initTableMapperJob > > I apologize for the spam. If this isn't the right way, please let me know. > Thanks. > > > --- On Wed, 10/14/09, Something Something <luckyguy2...@yahoo.com> wrote: > > From: Something Something <luckyguy2...@yahoo.com> > Subject: Question about MapReduce > To: general@hadoop.apache.org, hbase-u...@hadoop.apache.org > Date: Wednesday, October 14, 2009, 10:18 PM > > I would like to start a Map-Reduce job that does not read data from an > input file or from a database. I would like to pass 3 arguments to the > Mapper & Reducer to work on. Basically, these arguments are keys on the 3 > different tables on HBase. > > In other words, I don't want to use FileInputFormat or DbInputFormat > because everything I need is already on HBase. > > How can I do this? Please let me know. Thanks. > > > > > > > > > -- Pro Hadoop, a book to guide you from beginner to hadoop mastery, http://www.amazon.com/dp/1430219424?tag=jewlerymall www.prohadoopbook.com a community for Hadoop Professionals
