python multiprocessing Process is killed by http request if ipdb is imported

问题

It seems simply importing ipdb when making an http request wrapped in a multiprocessing Process instance causes the program to exit with no errors or messages.

The following script behaves very strangely:

from multiprocessing import Process
import requests
import ipdb

def spawn():

    print("before")
    r = requests.get("http://wtfismyip.com")
    print("after")

Process(target=spawn).start()

If you run this in terminal the output is simply before and you are back at your prompt. If you comment out import ipdb everything is fine and the request is successfully made.

• Storing the Process instance in a variable and calling join() after start() didn't make a difference.
• This happens in both Python 2.7.10 and 3.5.0.
• It does not happen with the traditional pdb.
• Other people here and here have also had this issue. In the former I am not sure if importing ipdb was the cause. In the latter it appeared to be an package/python version upgrade issue, but I checked that my iPython and ipdb are the current latest (4.0.0 and 0.8.1).

Can anyone explain why this is happening?

来源：https://stackoverflow.com/questions/33877491/python-multiprocessing-process-is-killed-by-http-request-if-ipdb-is-imported