This page will guide you through the process of making GILA up and running.
- Maven, needed to build GILA.
- A Hadoop 2.6.0 cluster. You can find more details on how to setup yours here.
- Hadoop counters must be increased. A value of at least 1500 is advised, but bigger graphs may need more. The value can be set by modifying mapred.job.counters.limit in mapred-site.xml Hadoop configuration file.
- The three running scripts: gila.sh, partition.sh, layout.sh.
- Clone the GIT repository on your machine.
- Navigate to the root folder of the cloned project.
Run the Maven build with the following command:
$ mvn clean package
- Once the build completes, go to gila-core -> target, the file gila-core-0.8-jar-with-dependencies.jar is the compiled version of GILA.
Download the running scripts and place them in the same folder, the same containing the GILA compiled jar. Run the following command:
$ sh gila.sh <path to input file> <path to output file> <number of workers>
Please note that both paths, input and output, must be on hdfs. Those scripts are just an example on how the jobs can be launched and can be easily customized.
In order to customize the layout or partitioning process, you can use the options listed in the options page.