python - TCP 服务端程序退出后，为什么端口还是一直被占用？

Question

服务端

import os
import socket
import sys
import time
import threading
from loguru import logger
from concurrent.futures import ThreadPoolExecutor
from concurrent.futures._base import Future
import multiprocessing

default_encoding: str = 'utf-8'

pool = ThreadPoolExecutor(
    max_workers=20,
    thread_name_prefix='simple-work-thread-pool'
)


def init_serversocket() -> socket.socket:
    serversocket = socket.socket(
        family=socket.AF_INET,
        type=socket.SOCK_STREAM
    )

    # 获取本地主机名
    host = socket.gethostname()

    logger.debug(f'host {host}')

    port = 6001

    # 绑定端口号
    serversocket.bind(('0.0.0.0', port))

    # 设置最大连接数，超过后排队
    serversocket.listen(2048)

    return serversocket


def send_response(clientsocket: socket.socket, addr: tuple, response_body: bytes) -> int:
    send_len: int = clientsocket.send(response_body)
    clientsocket.close()
    return send_len


def start_request(clientsocket: socket.socket, addr: tuple) -> int:
    try:
        pid = os.getpid()
        # logger.debug(f'pid: {pid}, get message from {addr}')
        request_body: bytes = clientsocket.recv(2048)
        request_text: str = request_body.decode(encoding=default_encoding)

        response_text: str = f'server get message: {request_text}'

        response_body: bytes = response_text.encode(default_encoding)
        # time.sleep(1)
        send_len = send_response(
            clientsocket=clientsocket, addr=addr, response_body=response_body)
        # logger.debug(f'发送了响应')
        return send_len
    except Exception as error:
        logger.exception(error)


def start_request_callback(future: Future) -> None:
    send_len: int = future.result()
    logger.debug(
        f'{threading.current_thread().name}, send payload len is {send_len}')


if __name__ == "__main__":
    serversocket = init_serversocket()

    pool = multiprocessing.Pool(processes=16)

    try:
        while True:
            clientsocket, addr = serversocket.accept()

            clientsocket: socket.socket
            addr: tuple

            # future: Future = pool.submit(start_request, clientsocket, addr)
            # future.add_done_callback(start_request_callback)

            pool.apply_async(start_request, (clientsocket, addr))
    finally:
        pool.close()
        pool.join()

        serversocket.close()

客户端：

from base64 import encode
import socket  # 客户端 发送一个数据，再接收一个数据
import json
from loguru import logger
from concurrent.futures import ThreadPoolExecutor
import time
import multiprocessing


failture_requests = []


def send_request(index: int):
    try:
        # 声明socket类型，同时生成链接对象
        clientsocket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
        clientsocket.connect(('127.0.0.1', 6001))  # 建立一个链接，连接到本地的6969端口

        payload = b'ponponon'

        clientsocket.send(payload)

        data = clientsocket.recv(1024)

        payload = data.decode()
        logger.debug(index)

        clientsocket.close()
        logger.debug('请求完成')
    except Exception as error:
        failture_requests.append(index)
        logger.exception(error)


if __name__ == "__main__":

    pool = multiprocessing.Pool(processes=32)

    for index in range(1000000):

        pool.apply_async(send_request, (index, ))

    pool.close()
    pool.join()

问题症状：

• 启动服务端后立刻关闭，再启动服务端也一切正常
• 启动服务端后，也启动客户端请求服务端，然后立刻关闭服务端。这个时候，启动就无法再次启动服务端，服务端会报错：OSError: [Errno 98] Address already in use
• 重点是，使用 sudo lsof -i :6001 空空如也，没有任何进程再占用这个 6001 端口
• 这个 OSError: [Errno 98] Address already in use 会在等待几分钟甚至几十分钟后自己消失，这个时候就能重新启动服务端了

为什么？

服务端的报错

╰─➤  python 002_socket_tcp_server_process.py                                                                                                  1 ↵
2023-05-04 23:00:42.862 | DEBUG    | __main__:init_serversocket:28 - host amd5700g
Traceback (most recent call last):
  File "/home/pon/Desktop/code/me/http_wsgi_asgi_tutorial/002_socket_tcp_server_process.py", line 73, in <module>
    serversocket = init_serversocket()
                   ^^^^^^^^^^^^^^^^^^^
  File "/home/pon/Desktop/code/me/http_wsgi_asgi_tutorial/002_socket_tcp_server_process.py", line 33, in init_serversocket
    serversocket.bind(('0.0.0.0', port))
OSError: [Errno 98] Address already in use

sudo lsof -i :6001 的输出, 空空如也

(http_wsgi_asgi_tutorial) ╭─pon@amd5700g ~/Desktop/code/me/http_wsgi_asgi_tutorial  ‹master*› 
╰─➤  sudo lsof -i :6001                                                                                                                                                     1 ↵
(http_wsgi_asgi_tutorial) ╭─pon@amd5700g ~/Desktop/code/me/http_wsgi_asgi_tutorial  ‹master*›

为什么？为什么？是 lsof 命令不准确吗?

---

换成 sudo netstat -anp | grep 6001 命令试了试，有输出

╰─➤  sudo netstat -anp | grep 6001                                                                                  130 ↵
tcp        0      0 127.0.0.1:6001          127.0.0.1:56470         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:38966         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:50820         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:60648         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:50472         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:50860         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:50040         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:44920         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:38078         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:54364         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:35426         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:54430         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:56028         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:58210         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:38254         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:45726         TIME_WAIT   -                   
tcp        0      0 127.0.0.1:6001          127.0.0.1:38046         TIME_WAIT   -                   
.... 后面的省略，实在是太多了               
             

Answer 1

这就是典型的 TIME_WAIT 端口占用的问题呗：程序意外退出，一个已连接的 Socket 没被主动释放，只能等操作系统最后回收。

解法即是侦听前加 SO_REUSEADDR 参数。

serversocket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
serversocket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1) # 在 bind 之前加

P.S.1. Linux 3.9 之后的内核又新增了一个 SO_REUSEPORT 参数，比 SO_REUSEADDR 来说更精细化了。不过一般来说的话都是同时设置两者的。Windows 上出于安全考虑一般还需要额外设置 SO_EXCLUSIVEADDRUSE。

P.S.2 lsof 扫的是进程，它实质看的是“某个文件被某个进程占用的情况”（Linux 下万物皆文件嘛，Socket 也是个文件）。你这进程都没了，连 pid 都没有当然就看不到了。