We use CDH5.4. I was able to change the hive-site.xml using the CM option - enable Hive on Spark. That same hive-site.xml has been copied to the Zeppelin conf folder. Am I missing anything? As I said, I can confirm that HOS works from inside Hive CLI (but that is different from using HS2).
On Fri, Jul 10, 2015 at 9:45 AM, John Omernik <j...@omernik.com> wrote: > Zeppelin is connecting to a Hive Thrift Server (Hiveserver2) in your org. > Can you run another instance where you start it with the execution engine > being spark or perhaps set it in the hive-site.xml? > > On Fri, Jul 10, 2015 at 10:51 AM, Abhi Basu <9000r...@gmail.com> wrote: > >> SET hive.execution.engine=spark within Zeppelin seems to have no effect. >> In Hive CLI, it does work. >> >> On Thu, Jul 9, 2015 at 7:20 PM, <felixcheun...@hotmail.com> wrote: >> >>> Could you set the execution engine using thesetConf method >>> on SQLContext or by running SET key=value commands using SQL statement? >>> >>> >>> >>> On Thu, Jul 9, 2015 at 7:09 AM -0700, "Abhi Basu" <9000r...@gmail.com> >>> wrote: >>> >>> How to use Hive on Spark from within Zeppelin? We are using Cdh5.4 >>> where we are able to enable HOS and I can enable this in hive cli by >>> setting hive.execution.engine=spark and see the differences in query times. >>> Queries in Zeppelin executes on MR instead. Hive configuration file in >>> Zeppelin is updated with latest version from Hive. >>> >>> Thanks >>> >>> Abhi >>> >>> >>> >>> Sent via the Samsung Galaxy Note®, an AT&T 4G LTE smartphone >>> >> >> >> >> -- >> Abhi Basu >> > > -- Abhi Basu
