I met with a strange behavior in one of my jobs. Attached is the job design, it extracts around 100 million records from one source and around 75 million records from another one, then trying to inner join them.
I was using Hash partitioning on the join keys and 'Stable Sort' option in the join stage itself, but data were not coming out from the join stage!! Then I replaced the stable sort option with a standalone sort Stage to enhance the sorting before join, but now data is not coming out of sort stage as well! (as you can see in job design image)
I know that the common thing between join & sort stages is the scratch file usages, the job already threw a scratch file error but it's gone now after increasing the scratch file space.
Job Design Image:
![Image](http://s11.postimg.org/pmbaoujtf/Job.jpg)
Appreciate your support.
Thanks