We have controlled the Hadoop jobs using the Perl API so far, which is quite limited.

The Hadoop itself uses many configuration options. Every option has a (dot-separated) name and a value and can be set on the command line using -Dname=value syntax:

perl script.pl run [-jt cluster_master | -c cluster_size [-w sec_to_wait]] [-r number_of_reducers] [Hadoop options] input_path output_path

Mind that the order of options matters – the -jt, -c, -w and -r must precede Hadoop options.