-
Notifications
You must be signed in to change notification settings - Fork 39
Home
This is the home page for the gluster file system plugin for hadoop. This plugin intends to be a drop in replacement for any applications which rely on the hadoop FileSystem for functionality (note - this is not limited to hadoop mapreduce, as there are many libraries, including apache HBase and Berkeley's Spark, which also rely on the FileSystem interface for distributed storage).
While Apache Hadoop typically ships with the Hadoop Distributed FileSystem (HDFS), Hadoop can be configured to use alternate FileSystems by leveraging its inherent pluggable filesystem architecture. One achieves this by building a plugin that acts as a mediator between the Hadoop FileSystem Interface and the desired FileSystem, and configuring Hadoop to use the plugin. This project contains the plugin required to enable GlusterFS to be used as a Hadoop FileSystem.
The Plugin supports both Apache Hadoop 1.x and 2.x releases and has thus far been successfully tested against both the Hortonworks and Intel Hadoop Distributions. The plugin is backwards compatible and thus we recommend you always obtain the latest version (the highest numerical value).
You can download a release directly from our maven repository.
Our Configuration page provides a guide on how to configure GlusterFS and various Hadoop ecosystem tools to make use of the plugin.
Our HowToContribute page contains information on how to participate in the development of the plugin.
Our HowToBuild page contains information on how to build the plugin if you would like to build your own plugin JARs. Building the plugin is simple: only maven and java are required.