Download or clone the Cubert repository (say, into /path/to/cubert).
> cd /path/to/cubert > ./gradlew > export CUBERT_HOME=/path/to/cubert/release
The gradlew command create a new folder called release which is what we will need to run cubert. This folder can be copied to hadoop cluster gateway.
To run cubert, first make sure that Hadoop is installed and the HADOOP_HOME environment variable points to the hadoop installation. Set the CUBERT_HOME environment to the release folder (note: CUBERT_HOME points to the release folder and not the “root” repository folder).
To verify that everything is in place, lets run the cubert command
> $CUBERT_HOME/bin/cubert -h Using HADOOP_CLASSPATH=:/path/to/cubert/release/lib/* usage: ScriptExecutor <cubert script file> [options] -c,--compile stop after compilation -D <property=value> use value for given property -d,--debug print debuging information ...
Cubert had been tested on both Hadoop versions 1.x and 2.x. You can set your version of hadoop in the gradle.properties file.
Running Cubert in Local Mode
Cubert can run in local mode as well. Although, we have to make sure that we have downloaded Hadoop distribution on the local computer and assigned the HADOOP_HOME environment variable.
Follow this steps to upgrade to the newest version of Cubert:
> cd /path/to/cubert > git pull > ./gradlew clean dist