Are you able to run the hive cli from the same installation? There are currently some issues while running metadata-only calls (show, describe) via JDBC. Regular queries should be fine though.
On 5/13/09 3:59 PM, "Aaron Kimball" <aa...@cloudera.com> wrote: > Hi all, > > I've been trying to use the Hive JDBC client today with some frustration. My > goal was to execute a simple "SHOW TABLES" statement in Hive. > > If I start the Hive server with HIVE_PORT=10000 hive --service hiveserver, the > following happens when I connect to jdbc:hive://localhost:10000/default: > java.sql.SQLException: Method not supported > > If instead I attempt to connect to jdbc:hive:// (without the standalone > hiveserver started), I get: > java.sql.SQLException: MetaException(message:hive.metastore.warehouse.dir is > not set in the config or blank) > > I'm confused where I should set the hive.metastore.warehouse.dir property. > I've run this from a directory containing a valid hive-site.xml; this > directory is named "conf/", so I also tried running my program in that > directory's parent, thinking it may look for conf/hive-default.xml and > conf/hive-site.xml. How do I set the configuration files that will be loaded > inside the call to DriverManager.getConnection()? And in the case of the > standalone server, does anyone have any insight into why I'd get "method not > supported" ? > > FWIW, the program I ran was invoked via 'hadoop jar ...'; I don't know if > launching a program in this way would mess up Hive's config paths, etc. This > is Hadoop 0.18.3, Hive 0.3.0. > > Thanks, > - Aaron
