使用Spark SQL 的JDBC server

使用 spark SQL的JDBC Server


修改 hive-site.xml 文件

<property>
  <name>hive.server2.thrift.bind.host</name>
  <value>192.168.1.100</value>
</property>

启动spark SQL的JDBC server

 $ cd  spark/
 $ ln -s ./hive/conf/hive-site.xml hive-site.xml

 $ chmod- -R 777 spark/logs/

 $ cd spark/sbin

# 启动 thriftserver
$ ./start-thriftserver.sh  --master yarn --hiveconf hive.server2.thrift.port=10008

或者这样启动

./start-thriftserver.sh  \
    --master yarn  \
    --conf spark.shuffle.service.enabled=true \
    --conf spark.dynamicAllocation.enabled=true \
    --conf spark.dynamicAllocation.minExecutors=1 \
    --conf spark.dynamicAllocation.maxExecutors=10 \
    --hiveconf hive.server2.thrift.port=10008

beeline 连接JDBCServer

[htestn@localhost bin]$ ./beeline -u jdbc:hive2://hadoop2.com:10008 -n htest
scan complete in 3ms
Connecting to jdbc:hive2://hadoop2.com:10008
Connected to: Spark SQL (version 1.4.0)
Driver: Spark Project Core (version 1.4.0)
Transaction isolation: TRANSACTION_REPEATABLE_READ
Beeline version 1.4.0 by Apache Hive
0: jdbc:hive2://hadoop2.com:10008> show databases;
+-----------+
|  result   |
+-----------+
| dp_tmp    |
+-----------+
1 rows selected (1.132 seconds)

注：
如果是thrift server 权限问题 在连接的时候指定hadoop可操作的用户 ./beeline -u jdbc:hive2://2.hadoop.com:10008 -n htest（hadoop 账户）