We do provide executing hive queries in Linkedin grids, which also works with secure hadoop, tested with hive-0.8.1
Essentially that is a jobtype that uses java job to run the HiveViaAzkaban.
You need to fill the ${hive.home} in both private.properties and plugin.properties.
You also need to place your udf jars in ${hive.aux.jars.path}.
If ${hadoop.home} is not set globally, you will need it set in hive job type as well.
Also make sure ${hive.home}/conf is configured correctly. This is especially when you evaluate with a fresh hive with a derby DB, it will default to a db file with hive bin, which likely is not where your azkaban job working directory is set to.