We are running HDP 3.1 (latest release as of Feb 2019).
We configured Spark to Hive+Ranger integration as instructed here:
This works OK from spark shell; eg. you can access Hive tables from Spark, and Ranger policies do apply correctly.
However, when you do the same from Zeppelin, Ranger seems to be out of the picture... you can access all Hive tables and rows as it Ranger would not exist.