Beyond MapReduce - Enterprise Data Workflows with Cascading - page 86

$ export PATH = ~/.lingual-client/bin/: $PATH
When using Lingual with Apache Hadoop, the SQL command shell expects certain
environment variables to be set. That way the correct Hadoop version and configuration
will be included in the CLASSPATH :
HADOOP_HOME
Path to local Hadoop installation
HADOOP_CONF_DIR
Defaults to $HADOOP_HOME/conf
HADOOP_USER_NAME
The username to use when submitting Hadoop jobs
Assuming that you have HADOOP_HOME already set, then:
$ export HADOOP_CONF_DIR = $HADOOP_HOME /conf
$ export HADOOP_USER_NAME = <username>
If you're working with a remote Elastic MapReduce cluster on Amazon AWS, see the
Bash EMR utilities. Specifically, use the emrconf command to fetch remote configura‐
tion files.
If you encounter errors executing SQL queries on a remote cluster (Amazon AWS,
Windows Azure HDInsight, etc.) try the following workaround:
$ export HADOOP_USER_NAME = hadoop
That should resolve security issues that may be causing failures on the remote cluster.
Now let's try using the Lingual SQL command shell. The following example is based on
data from the MySQL Sample Employee Database :
$ mkdir -p ~/src/lingual
$ cd ~/src/lingual
$ curl http://data.cascading.org/employees.tgz | tar xvz
That creates an employees subdirectory for the table data, which is essentially several
large CSV files. Next, load the schema for these tables into Lingual using SQL data
definitions:
$ curl http://data.cascading.org/create-employees.sh > create-employees.sh
$ chmod +x ./create-employees.sh
$ ./create-employees.sh local
Now try the SQL command line, querying to show a relational catalog for these tables:
$ lingual shell
0: jdbc:lingual:local> !tables
That lists metadata about the available tables: EMPLOYEE , TITLES , SALARIES . Next, let's
try a simple query: