连接到 Multiprocessing BaseManager 在本地工作，但不适用于远程服务器

Question

I'm currently using themultiprocessing.managers.BaseManager functionality to serve some Python objects so that I could interact with their proxies on the client side.

After I start the server defined by the BaseManager, I try to connect at the client side to access the serveables. However, I'm seeing an EoFError only when the server and client exist on different hosts. If they were on the same local host, I can get access just fine.

server.py

from multiprocessing.managers import BaseManager
password = "password".encode("utf-8")

class MyManager(BaseManager):
    pass

a = {'one': 1}  # to be served
MyManager.register("get_a", callable=lambda: a)
m = MyManager(address=("127.0.0.1", 60000), authkey=password)
m.start()
# ... long running process ...

client.py

from multiprocessing.managers import BaseManager
password = "password".encode("utf-8")

class MyManager(BaseManager):
    pass

MyManager.register("get_a")
manager = MyManager(address=("SERVER_IP", 60000), authkey=password)
manager.connect()  # raises EoFError

Traceback:

File "client.py", line 20, in connect_to_executor_manager
               manager.connect()
File "/usr/lib/python3.7/multiprocessing/managers.py", line 512, in connect
              conn = Client(self._address, authkey=self._authkey)
File "/usr/lib/python3.7/multiprocessing/connection.py", line 498, in Client
            answer_challenge(c, authkey)
File "/usr/lib/python3.7/multiprocessing/connection.py", line 741, in answer_challenge
              message = connection.recv_bytes(256)         # reject large message
File "/usr/lib/python3.7/multiprocessing/connection.py", line 216, in recv_bytes
            buf = self._recv_bytes(maxlength)
File "/usr/lib/python3.7/multiprocessing/connection.py", line 407, in _recv_bytes
          buf = self._recv(4)
File "/usr/lib/python3.7/multiprocessing/connection.py", line 383, in _recv
          raise EOFError

Answer 1

I just ran into the same problem, but luckily, I created a clue when I modified my code in the course of debugging

from multiprocessing.managers import BaseManager
password = "password".encode("utf-8")

class MyManager(BaseManager):
    pass

a = {'one': 1}  # to be served
MyManager.register("get_a", callable=lambda: a)
m = MyManager(address=("127.0.0.1", 60000), authkey=password)
print(f"serving at address {m.address}")      # I added the equivalent of this line
m.start()

My test code caused the "serving at address..." line to print twice. This made me think that the code was getting spawned twice. I recall seeing somewhere that the server code should be contained in an if __name__ ==... block, so I tried that and things started working.
I'm no expert, but I think this error is caused during the spawn/forking/etc. process. Long story short, you can modify your server.py code like so and it will likely work.

from multiprocessing.managers import BaseManager
password = "password".encode("utf-8")

class MyManager(BaseManager):
    pass

a = {'one': 1}  # to be served
MyManager.register("get_a", callable=lambda: a)

if __name__ == '__main__':
    m = MyManager(address=("127.0.0.1", 60000), authkey=password)
    m.start()