response

Response to J H (too much text to fit into a comment)

buffer size

The Hub buffer size is 400,000 so that the buffer is not blocking the chunk data that is coming in in any way. There are typically 150-300 chunks in a transfer but as far as i know the buffer is overridden by the newest chunk as data comes in recieved = socket.Receive(buffer); since I process chunks serialy.

Logging

On average it takes 2-4 seconds to transfer a chunk. If I put another writeline in the RecieveData of the interconnect helper I can see that it there appears to be some sort of limit to the amount of data that can be read from one socket.receive.

logging in the interconnect

while (filled < length)
{
    if (!Helpers.NetworkHelpers.Connected(socket)) throw new SocketDisconnectedException();
    recieved = socket.Receive(buffer);
    Array.Copy(buffer, 0, output, filled, recieved);
    filled += recieved;
    Console.WriteLine("Interconnect. " + filled + " of " + length);
}
logging in the chunk receiver

for(int i = 0; i < chunkAmount; i++)
{
    socket.Send(InterconnectHelper.FormatSendData(Encoding.ASCII.GetBytes(i.ToString())));
    Console.WriteLine("Chunk " + i + " of " + (chunkAmount -1));
    received = socket.Receive(buffer);
    recievedData = InterconnectHelper.RecieveData(buffer, received, socket);

    Array.Copy(recievedData, 0, returnData, i * chunkSize, recievedData.Length);
}
Paste bin of logs with default sizes Paste bin of logs with 200k chunksize

swap memory

The memory isn't a problem on either end. On the Pi3 there is still ~630MB (im assuming that the vmstat 1 shows KB) and ~160MB on the Pi Zero. Both of them show 0 in the swap columns as the process runs so that is not a bottleneck.