This project was created as a self-challenge assignment for a course I took.

Building

meson build
cd ./build
meson compile

Running

<Within build dir>
To start the Swarm server <By default binds to all interfaces on port 50001>:
./DiscoveryServer
To start a Peer<By default expects the Swarm server to be on 127.0.0.1:50001>:
./Peer

Overview

The swarm server is started first. When peers start up, they connect to the swarm server, and give it their (ip address, port) pair that they are listening to client connections on.

Swarm Server Header:
Mode: 1, // Add or remove peer from client or swarm server
address: 4, // The IPv4 address of the peer
port: 2 // the port of the peer
Total Header Length: 7

This connection to the swarm server is kept open for the entire amount of time the peer is participating in the swarm managed by this swarm server. When the peer disconnects from the Swarm server, it is removed from the swarm.

The way that a peer can download a file, is by sending a file request message to all other participating peers in the swarm. This message contains the filename of the file that the client is looking for.

File Request -> Request a File from another client.
    Populated header fields:
        - message_type: 0
        - requested_filename

These clients send back a File response message indicating whether they have it or not. If they do have it, they will tell the client how many chunks the file takes up as well. (chunks are 32 megabytes each [defined in ApplicationLayer/Peer.hpp])

File Response -> Respond to a file request whether we have it.
    Populated header fields:
        - message_type: 1
        - num_chunks (0 if they don't have it)

The peer looking for a file will now send a file request message to each of the clients that have the file. Requesting a roughly equal amount of chunks from each of the other peers.

For example, if there were 4 peers with the file, 1/4 of the chunks would be requested from each peer.

Chunk Request -> We found out that a client has a file we want, request a set of chunks from that client.
    Populated header fields:
        - requested_filename
        - chunk_request_begin_idx
        - chunk_request_end_idx

After which, each of the peers will respond with the requested chunks like so:

Chunk Response -> Respond to the client with the chunks that they requested.
    Populated header fields:
        - chunk_idx
        - current_chunk_size
        - contains 32Mb payload after header

The chunks will be written in parallel to to specified download location with the filename(s) <chunk_num>.p2p.

Then they will be re-merged into a file under the same name as the peer it was downloaded from.