Search code examples
pythonwindowssocketsflaskipv6

Python flask dual-stack on WINDOWS (ipv4 and ipv6)


I'm using python Flask on Windows 10. It works fine for ipv4 or ipv6, depending on the ip I bind, but not both at the same time.

With this example:

from flask import Flask

app = Flask(__name__)

@app.route("/")
def hello_world():
    return "Hello World! <strong>I am learning Flask</strong>", 200

app.run(host='', port=5000, debug=True)

I get this

With this example:

from flask import Flask

app = Flask(__name__)

@app.route("/")
def hello_world():
    return "Hello World! <strong>I am learning Flask</strong>", 200

app.run(host='::', port=5000, debug=True)

I get this

When I run a minecraft server and bind it to "::", I get this

I saw in this post that Flask does dual-stacking on Linux when binding to "::". I would like to know if it's possible to make de Flask app listen to both ipv4 and ipv6 address at the same time. Thanks.

Additional info: Python socket module does not dual stack on windows (i think it does on Linux). I found this and managed to create a dual stack socket with this example:

import socket
from recipe1 import has_dual_stack, create_server_sock, MultipleSocketsListener
tcp = MultipleSocketsListener([("0.0.0.0", 5000), ("::", 5000)])
while True:
    con, cliente = tcp.accept()
    print ('Concetado por', cliente)
    while True:
        msg = con.recv(1024)
        if not msg: break
        print (cliente, msg)
    print ('Finalizando conexao do cliente', cliente)
    con.close()

result It does work, but I don't know if this socket can be used with Flask.


Solution

  • According to [PalletProjects.Flask]: run(host=None, port=None, debug=None, load_dotenv=True, **options):

    Do not use run() in a production setting. It is not intended to meet security and performance requirements for a production server. Instead, see Deployment Options for WSGI server recommendations.

    NGINX knows how to handle this exact scenario.

    In development mode, I don't know why is so important to listen on all addresses, as things can be tested listening on one at a time.

    I didn't see any easy way of making this work. As a note, on Lnx things seem to be easier as the IPv4-mapped IPv6 addresses are controlled via the net.ipv6.bindv6onl setting.

    There are however a number of ways to work things around, here's one that executes the current file (itself) in a new process for each listening IP address (and does it in a thread, since the (child) process blocks the execution).

    code00.py:

    #!/usr/bin/env python3
    
    import sys
    from flask import Flask
    import threading
    import subprocess
    
    
    app = Flask(__name__)
    
    
    def run_flask(host):
        return subprocess.call([sys.executable, sys.argv[0], host])
    
    
    @app.route("/")
    def hello_world():
        return "Hello World! <strong>I am learning Flask</strong>", 200
    
    
    def main(argv):
        port = 5000
        debug = True
    
        if argv:
            app.run(host=argv[0], port=port, debug=debug)
        else:
            hosts = [
                "127.0.0.1",
                "::1",
            ]
    
            threads = list()
            for host in hosts:
                threads.append(threading.Thread(target=run_flask, args=(host,)))
    
            for idx, thread in enumerate(threads):
                print("Starting on {0:s}:{1:d}".format(hosts[idx], port))
                thread.start()
    
    
    if __name__ == "__main__":
        print("Python {0:s} {1:d}bit on {2:s}\n".format(" ".join(item.strip() for item in sys.version.split("\n")), 64 if sys.maxsize > 0x100000000 else 32, sys.platform))
        main(sys.argv[1:])
        print("\nDone.")
    

    Output (it's a bit mixed):

    [cfati@CFATI-5510-0:e:\Work\Dev\StackOverflow\q057881991]> "e:\Work\Dev\VEnvs\py_064_03.07.03_test0\Scripts\python.exe" code00.py
    Python 3.7.3 (v3.7.3:ef4ec6ed12, Mar 25 2019, 22:22:05) [MSC v.1916 64 bit (AMD64)] 64bit on win32
    
    Starting on 127.0.0.1:5000
    Starting on ::1:5000
    
    Done.
    Python 3.7.3 (v3.7.3:ef4ec6ed12, Mar 25 2019, 22:22:05) [MSC v.1916 64 bit (AMD64)] 64bit on win32
    
    Python 3.7.3 (v3.7.3:ef4ec6ed12, Mar 25 2019, 22:22:05) [MSC v.1916 64 bit (AMD64)] 64bit on win32
    
     * Serving Flask app "code00" (lazy loading)
     * Serving Flask app "code00" (lazy loading)
     * Environment: production
     * Environment: production
       WARNING: Do not use the development server in a production environment.   WARNING: Do not use the development server in a production environment.
       Use a production WSGI server instead.
    
     * Debug mode: on
       Use a production WSGI server instead.
     * Debug mode: on
     * Restarting with stat
     * Restarting with stat
    Python 3.7.3 (v3.7.3:ef4ec6ed12, Mar 25 2019, 22:22:05) [MSC v.1916 64 bit (AMD64)] 64bit on win32
    Python 3.7.3 (v3.7.3:ef4ec6ed12, Mar 25 2019, 22:22:05) [MSC v.1916 64 bit (AMD64)] 64bit on win32
    
    
     * Debugger is active!
     * Debugger is active!
     * Debugger PIN: 566-002-078
     * Debugger PIN: 566-002-078
     * Running on http://127.0.0.1:5000/ (Press CTRL+C to quit)
     * Running on http://[::1]:5000/ (Press CTRL+C to quit)
    

    As seen, the servers start listening on the given addresses (you can remove the print calls, in order to have less output). Also (on another cmd):

    [cfati@CFATI-5510-0:C:\WINDOWS\system32]> netstat -an | findstr 5000
      TCP    127.0.0.1:5000         0.0.0.0:0              LISTENING
      TCP    [::1]:5000             [::]:0                 LISTENING
    

    You could also operate at OS level, by playing with the /etc/hosts file, but I didn't test that.