ThreadPoolExecutor中socket.bind()未触发‘Address already in use’报错问题求助

我在用Python编写基于socket的程序，使用ThreadPoolExecutor测试重复监听同一端口时遇到异常情况，以下是复现问题的最小示例代码：

import socket
import time
import concurrent.futures

def listen6666():
    server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
    server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
    print("set socket")
    server_socket.bind(("localhost", 6666))
    server_socket.listen()
    print("done")

# 主进程先绑定并监听6666端口
server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
server_socket.bind(("localhost", 6666))
server_socket.listen()

with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
    fn = executor.submit(listen6666)
    fn2 = executor.submit(listen6666)
    # 直接调用listen6666()会触发"Address already in use"错误

# 取消注释下面两行会触发错误
# fn.result() - "Address already in use"
# fn2.result() - "Address already in use"

time.sleep(10)

运行程序后无预期报错，仅输出：

$ python test.py 
set socket
set socket
$

我预期listen6666函数中的server_socket.bind会触发"Address already in use"错误，但只有直接调用listen6666，或者在sleep前调用fn.result()时才会报错。希望在ThreadPoolExecutor的线程中调用listen6666时也能触发该报错，该如何解决？

问题核心在于ThreadPoolExecutor的异步执行机制：用executor.submit()提交任务后，线程会后台异步执行，任务抛出的异常不会主动暴露，必须主动获取Future对象的结果或异常才会触发报错。

有两种可行的解决方法：

1. 主动获取任务结果并捕获异常
   在提交任务后，调用Future对象的result()方法，并通过try-except捕获异常，这样就能及时触发并显示错误：

with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
    fn = executor.submit(listen6666)
    fn2 = executor.submit(listen6666)
    try:
        fn.result()
    except Exception as e:
        print(f"任务1报错: {e}")
    try:
        fn2.result()
    except Exception as e:
        print(f"任务2报错: {e}")

2. 使用executor.map()替代submit()
map()方法会自动迭代任务执行结果，当任务抛出异常时，迭代到对应任务的位置就会触发异常：

with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
    # 提交两次listen6666任务
    results = executor.map(listen6666, [None, None])
    try:
        for _ in results:
            pass
    except Exception as e:
        print(f"任务报错: {e}")

补充说明：你设置的SO_REUSEADDR选项仅允许端口在TIME_WAIT状态下被复用，而当前主进程已绑定并监听6666端口，此时其他线程再绑定同一端口，即使开启该选项也会触发"Address already in use"错误，这是符合预期的行为。

内容的提问来源于stack exchange，提问作者turgenev