spark sql cli

1、将/letv/data/apache-hive-0.13.1-bin/conf/hive-site.xml、
/letv/data/apache-hive-0.13.1-bin/conf/ hive-log4j.properties.template
配置文件拷贝到$SPARK_HOME/conf目录下；

 cp /letv/data/apache-hive-0.13.1-bin/conf/hive-site.xml /letv/data/spark-1.5.0-bin-hadoop2.6/conf

将hive-site.xml中的jdbc的地址修改为有mysql的服务器地址

2、在/letv/data/spark-1.5.0-bin-hadoop2.6/conf/spark-env.sh下配置spark classpath ,添加mysql驱动类(在这之前要将mysql的jar放在lib目录下)

export SPARK_CLASSPATH=$SPARK_LOCAL_DIRS/lib/mysql-connector-java-5.1.18.jar

3、启动hive metastore

$nohup hive --service metastore > metastore.log 2>&1 &

4、修改/tmp/hive-root权限

hadoop fs -chmod -R +w /tmp/hive-root

5、启动 spark-sql cli

bin/spark-sql --master  spark://ip:7077 yarn-client

6、关闭info

log4j.properties.template下修改log4j.rootCategory=WARN, console

参考：
http://lxw1234.com/archives/2015/06/294.htm

java.lang.IllegalArgumentException: java.net.UnknownHostException: ns1

需要在spark-env.sh 中配置
export HADOOP_CONF_DIR=$HADOOP_HOME/etc/hadoop
是为了使用hdfs