Hue+mapreduce job configuration and problems encountered

Recently in the Hadoop project, in order to give customers a UI interface, the choice of hue, first use or encountered a lot of problems:

Let's talk about the configuration job properties when submitting a mapreduce job.

Because all jobs on hue, whether it is mapreduce, pig, etc. are dispatched through Oozie. And because of the oozie mechanism, it can't possibly go to the jar of your packaged MapReduce job to read your source code.

The configuration of a series of jobs that we configured in the main () function of MapReduce, Oozie was impossible to get, so he also reported all sorts of errors, and at the beginning I only had the input and output paths of the job.

Mapreduce.output.fileoutputformat.outputdir	Output path of the job
Mapreduce.input.fileinputformat.inputdir	Input path to the job

The map and reduce processes are not executed at all at the end of the job.

Later went to the official website to find the literature, see their demo, know the need to configure all the properties in the main function, so with all the attributes, and finally the error

On the internet to check a lot of information, all said because the map output format is problematic, but everyone is running in the command line job, and did not encounter on the hue, a temporary problem into an unresolved situation. Later, I think that since the properties of the job in the Hue page has been configured, the source code should not need the main function to configure, try to delete the main function and then recompile the package upload submission, all OK. The problem is solved.