Can you provide more details on what you are trying to do? All references to files/directories in a job specification refer to files/directories on HDFS.
Assuming you are referring to the job specification under Job Designer:
- mapred.input.dir /user/hue/jobsub/sample_data on hdfs
- $output_directory interpreted as a parameter that is prompted for when you submit the job. This directory should have write permissions for the current user. In this case sandbox.
As an example use:
- Go to Job Designer
- Clone streaming _wordcount
- Change the name of job to say my_streaming_wordcount (or any other name of your choice)
- Click Save at the bottom of this screen
- Once you are back on the job designer page, select my_streaming_wordcount
- Click Submit and it will prompt you for the output directory
- type /tmp/output1 for example
When the job is finished, you can go to this directory on the 'File Browser' and see results
Hope this helps....