I am using parallel stream to process large dataset. But it gives inconsistent results. The database used is Postgres. I have hierarchical data with levels defined. For exam