ChunkSafe: Content-Addressable Peer-to-Peer Backup for Trusted Networks Ben Lipton and Jonathan Gluck CPSC 87: Parallel and Distributed Systems Swarthmore College, Spring 2010

Included files: chunksafe.py: main program, call as follows: 1. Create first node chunksafe.py <backup time (HH MM)> 2. Create subsequent node chunksafe.py <backup time (HH MM)> 3. Resume killed node chunksafe.py <port to run on (must be the same as original node)>

startchunksafe.sh: ugly shell script for running program on Swarthmore network. Can be modified to work on other systems, but designed only to facilitate testing.

retrieve.py: file retrieval script, call as follows: retrieve.py <port of local daemon (if using startchunksafe.sh always 4000)>

bootstrap.py: program used for testing, creates 15 nodes on a single machine chunkfile.py: class abstracting a file on the filesystem so that we can get its stats easily chunks.py: classes related to making chunks out of a backup directory comm.py: DHT node used by ChunkSafeNode to communicate between machines fetcher.py: Class that uses Twisted to download a file from file server server.py: Class that uses Twisted to send a file when requested via TCP

Directories: entangled: DHT implementation based on Twisted engine examples: Examples taken from Entangled for reference googlevoice: Allows the program to send Jon a text message when the backup finishes (just for fun)