This is an attempt to recreate a resilient distributed file-system like Apache's HDFS. This was undertaken to try to better understand both a network-based file-system, and distributed-networks as a whole.

In order to achieve its resiliency, the file-system utilizes both regular heartbeats to detect failures, and Reed-Solomon code algorithms to repair corruptions upon detection.

It should be noted that this was developed locally, without access to a network cluster. The system still leverages TCP and an event-driven API to communicate between logically independent nodes. The 'Distributed' part is simulated, with each node having its own local filesystem, given via a relative home path specified in the 'chunkServers' config file. A minimum (if any, really) amount of changes should be necessary to convert this to a true local cluster.

As a side note, the name comes combining an awesome professor who taught me about all of this stuff during my last year in college, with the software this repo is attempting to replicate some features of (HDFS). A minimum amount of 3rd party libraries were used, and only to simplify extra stuff like terminal auto-complete, and string formatting.

• Java 8+
• Linux or Unix-Like (Cygwin, MinGW, etc) terminal with Tmux installed
• Maven

Note: Any operating system should work given that this is Java.

1. Clone the Repository
2. While inside the Distributed_File_System directory, run maven compile to compile and bundle into a fat jar inside of ./target
3. Edit chunkServers_[OS]. File line format: [serverName] [Port] [dynamicPath]
   Note: The start-up script terminal_setup.sh uses this file
4. java -cp /target/Distributed_File_System-1.0.jar fileSystem.node.Controller [PORT]