How to count connected clients in TCPServer?

Question

I'm using Pythons SocketServer.ThreadingTCPServer. Now I want to know how many clients are connected at a certain moment.

How to solve this?

Answer 1

SocketServer.ThreadingTCPServer swap a new thread for each client connection so knowing the number of client connected in a certain moment is the same as knowing how many thread are alive in that moment , so just use threading.activeCount and the number of client will be:

num_client = threading.activeCount() - 1 # Don't count the main thread.

---

Of course this will not give you correct result if there is other part of your code that swap thread too, so to fix that you can overwrite the process_request() and process_request_thread() methods by adding a counter for clients.

From the example here i wrote this snippet of code to test the two methods

import time
import socket
import threading
from SocketServer import ThreadingTCPServer, BaseRequestHandler


def client(ip, port, message):
    sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
    sock.connect((ip, port))
    sock.send(message)
    response = sock.recv(1024)
    sock.close()


class ThreadedTCPRequestHandler(BaseRequestHandler):

    def handle(self):
        data = self.request.recv(1024)
        cur_thread = threading.currentThread()
        response = "%s: %s" % (cur_thread.getName(), data)
        self.request.send(response)
        time.sleep(1)


class MyServer(ThreadingTCPServer):

    def __init__(self, *args, **kws):
        self._num_client = 0
        ThreadingTCPServer.__init__(self, *args, **kws)

    def process_request(self, *args, **kws):
        print "swap thread"
        self._num_client += 1
        ThreadingTCPServer.process_request(self, *args, **kws)

    def process_request_thread(self, *args, **kws):
        ThreadingTCPServer.process_request_thread(self, *args, **kws)
        print "kill thread"
        self._num_client -= 1

    def get_client_number(self):
        return self._num_client


def my_client_count(ignore=1):
    return  threading.activeCount() - ignore


if __name__ == '__main__':
    server = MyServer(("localhost", 0), ThreadedTCPRequestHandler)

    server_thread = threading.Thread(target=server.serve_forever)
    ip, port = server.server_address

    server_thread.setDaemon(True)
    server_thread.start()

    print "client 1 connected"
    client(ip, port, "Hello World 1")
    print "number of client get_client_number : %s,  enumerate : %s" \
       %  (server.get_client_number(), my_client_count())
    print "client 2 connected"
    client(ip, port, "Hello World 2")
    print "number of client get_client_number : %s,  enumerate : %s" \
       %  (server.get_client_number(), my_client_count())

    time.sleep(3)
    print "client 3 connected"
    client(ip, port, "Hello World 3")
    print "number of client get_client_number : %s,  enumerate : %s" \
       %  (server.get_client_number(), my_client_count())

Output :

client 1 connected
swap client thread
number of client get_client_number : 1,  enumerate : 2
client 2 connected
swap client thread
number of client get_client_number : 2,  enumerate : 3
kill client thread
kill client thread
client 3 connected
swap client thread
number of client get_client_number : 1,  enumerate : 2

Well as you can see the second way give more precise value, and the difference between the two ways is that because my server is running using a thread so i have always +1 thread that explain the difference of 1.

Well hope this can help :)

Answer 2

In a thread that "serves" client use some global count that is increased when client connects and decreased when disconnects.

If you want to count from OS level then use nestat -an with proper grep filter and wc -l (on Windows uses ports of grep and wc )