aiowstunnel

Persistent and reliable TCP tunneling based on asyncio.

Concepts

To illustrate the two differenc modes Client can operate in, we give a detailed example of two conceptually different setups.

Local Port Forwarding: the `CONNECT` mode

Let's suppose you have an application server listening on host srv:6789, you have two servers host1 and host2 and a client workstation cli. Let's suppose we have the following setup:

cli ---> host1 ---> host2 ---> srv

The arrow a ---> b indicates that a can connect to b, but b can not connect to a.

In this case you start a tunnel server on host2 that listens on port 9000.

Server('0.0.0.0', 9000)

You also strart a tunnel client on host1 in CONNECT mode with the following parameters

Client(
    aiowstunnel.CONNECT,  # you need connect mode here
    'host2', 9000,        # the tunnel server is listening on port 9000
    '0.0.0.0', 5678,      # the listening host/port
    'srv', 6789,          # the connectiong host/port
)

In this case, the parameters mean the following:

server_mode: This was set to aiowstunnel.CONNECT, which means the client wants the server to use the connect_host, connect_port settings when a new connection is received by the client and connect to the given host-port.
tunnel_host, tunnel_port: It tells the Client where the tunnel server is listening. The client will connect to this server and tries to keep the connection open.
listen_host, listen_port: The server_mode parameter is CONNECT, so the server is going to use the connect_host and connect_port parameters and the client will listen on this host/port. In this example the client is running on host1 and it will start a new server on 0.0.0.0:5678. The program running on cli can connect to this server as if it was the application server
connect_host, connect_port: As described earlier in CONNECT mode the server will use these to connect when a connection comes to the client. The client tells the server to connect to srv:6789 -- the application server.

Remote Port Forwarding: the `LISTEN` mode

What if the firewall is configured in this way:

cli ---> host1 <--- host2 ---> srv

In this case the tunnel server will be started on host1, the client on host2. The server needs to listen 0.0.0.0:5678 while the client connects to the application server.

Server('0.0.0.0', 9000)

Client(
    aiowstunnel.LISTEN,   # you need the server to listen
    'host1', 9000,        # the tunnel server is listening on port 9000
    '0.0.0.0', 5678,      # the server listens here
    'srv', 6789,          # the client connects here
)

Strarting the Server and the Client

Here is the code to properly start the server in the above example

import asyncio
import signal
import logging

from aiowstunnel.server import Server


logging.basicConfig(level=logging.INFO)


async def serve(stop):
    srv = Server('0.0.0.0', 9000)
    srv.start()
    await stop
    await srv.close()


loop = asyncio.get_event_loop()

# install signal handler
stop = asyncio.Future()
loop.add_signal_handler(signal.SIGINT, stop.set_result, None)

loop.run_until_complete(serve(stop))

Starting the client is very similar

import asyncio
import logging
import signal

from aiowstunnel.client import Client
import aiowstunnel


logging.basicConfig(level=logging.INFO)


async def provide_tunnel(stop):
    cli = Client(
        aiowstunnel.LISTEN,   # you need the server to listen
        'host1', 9000,        # the tunnel server is listening on port 9000
        '0.0.0.0', 5678,      # the server listens here
        'srv', 6789,          # the client connects here
    )
    cli.start()
    await stop
    await cli.close()


loop = asyncio.get_event_loop()

# install signal handler
stop = asyncio.Future()
loop.add_signal_handler(signal.SIGINT, stop.set_result, None)

loop.run_until_complete(provide_tunnel(stop))

Using SSL/TLS

TODO

richardbann/aiowstunnel