Open mtdeguzis opened 5 years ago
djmartin6
commented
5 years ago To add more info to this ticket. It seems that there is a workaround. When running set hive.execution.engine=mr; it fails but if we do set hive.hive.execution.engine=mr; it sets the correct value almost as if it's removing the hive. at the beginning of the parameter before running the query.
snuyanzin
commented
5 years ago I wonder why are you using sqlline version 1.5.0-SNAPSHOT
I tried the similar case with sqlline version 1.6.0, HDP-2.6.5, jdbc driver hive-jdbc-1.2.1000.2.6.5.0-292-standalone.jar but with much simpler connection string
like
sqlline> !connect jdbc:hive2://localhost:10000/default "" "" ""
Dec 27, 2018 10:31:44 AM org.apache.hive.jdbc.Utils parseURL
INFO: Supplied authorities: localhost:10000
Dec 27, 2018 10:31:44 AM org.apache.hive.jdbc.Utils parseURL
INFO: Resolved authority: localhost:10000
Transaction isolation level TRANSACTION_REPEATABLE_READ is not supported. Default (TRANSACTION_NONE) will be used instead.
0: jdbc:hive2://localhost:10000/default> !set version
version sqlline version 1.6.0
0: jdbc:hive2://localhost:10000/default> I tried mentioned commands for setting and viewing hive variables
set hive.execution.engine=mr;
set hive.execution.engine=tez;
set;and all of them look ok (see output below)
Are you sure you need to use sqlline version 1.5.0-SNAPSHOT rather than stable?
Are you sure you do not have the similar issue while working with beeline (I do not know but may be properties from your connection string could play some role here)
the output of mentioned commands in case of simple connection string (the output of set is cut)
0: jdbc:hive2://localhost:10000/default> set;
+-----+
| set |
+-----+
| _hive.hdfs.session.path=/tmp/hive/hive/7a831ab3-0a15-4afd-83c8-50a75d842e82 |
| _hive.local.session.path=/tmp/hive/7a831ab3-0a15-4afd-83c8-50a75d842e82 |
| _hive.tmp_table_space=/tmp/hive/hive/7a831ab3-0a15-4afd-83c8-50a75d842e82/_tmp_space.db |
| ambari.hive.db.schema.name=hive |
| atlas.cluster.name=Sandbox |
| atlas.hook.hive.maxThreads=1 |
| atlas.hook.hive.minThreads=1 |
| atlas.rest.address=http://sandbox-hdp.hortonworks.com:21000 |
...
| system:user.home=/home/hive |
| system:user.language=en |
| system:user.name=hive |
| system:user.timezone=UTC |
+-----+
984 rows selected (0.201 seconds)
0: jdbc:hive2://localhost:10000/default> set hive.execution.engine=tez;
No rows affected (0.002 seconds)
0: jdbc:hive2://localhost:10000/default> set hive.execution.engine=mr;
No rows affected (0.003 seconds)
0: jdbc:hive2://localhost:10000/default>
Works via normal on-cluster with Hortonworks Hadoop beeline, so I assume I am missing something simple...
beeline:
sqlline:
The property is listed with a normal set command under sqlline:
Seems we can't source the whitelist from the hiveserver:
... When that normally has many properties listed.