Spark standalone mode uses a simple built-in cluster manager. The Spark standalone mode cluster manager can also be run on a single machine, and it is great way to try out Spark (http://spark.apache.org/docs/latest/spark-standalone.html). In Standalone mode, Spark requires Java and Git installed on the system. Let's install the dependencies. On Ubuntu, the following command will install both Java and Git:
ubuntu@master:~$ sudo apt-get install openjdk-7-jdk git
The following are the steps to install Spark in standalone mode:
Download the latest Spark tarball from http://spark.apache.org/downloads.html. Spark packages are prebuilt compatible with a specific version of Hadoop and Mesos. At the time of writing, the latest version is 1.2, which is compatible with 0.18 version of Mesos and many versions of Hadoop, including the latest version, 2.4.
Extract the downloaded tar file and go to the following directory:
ubuntu@master:~$ tar –xzf spark-*.tar.gz ubuntu@master:~$ cd...