[Python-Dev] test_multiprocessing: test_listener_client flakiness

Wed Jun 18 18:48:20 CEST 2008

Hello,
2008年6月18日 Jean-Paul Calderone <exarkun at divmod.com>:
> On 2008年6月18日 18:35:26 +0200, Amaury Forgeot d'Arc
> <amauryfa at gmail.com> wrote:
>>>> [snip]
>>>> I just found the cause of the problem ten minutes ago:
>> It seems that when a socket listens on the address "127.0.0.1" or
>> "localhost", another process cannot connect to it using the machine's
>> name (even from the same machine).
>> That's because when you try to connect to A:B you won't connect to a
> server listening on X:B - somewhat by design. Changing the test to
> listen on A:B and X:B might fix it, but so would changing it to connect
> to the same address it listens to.
>>> The best seems to listen with the empty address "".
>> This will cause it to listen on all available interfaces, including
> public ones. It's rather unlikely that someone from the internet
> will connect to the port while the test suite is running and use it
> to do terrible things to you, but it's not /impossible/.
>> I'd suggest changing the client to connect to "127.0.0.1" or "localhost",
> instead.

OK, I am not a tcp expert.
How about the following patch?
It seems to me that it is wrong for the server to lie about the
address it listens to.
Index: ../../Lib/multiprocessing/connection.py
===================================================================

--- ../../Lib/multiprocessing/connection.py (revision 64378)
+++ ../../Lib/multiprocessing/connection.py (working copy)
@@ -216,8 +216,6 @@
 self._socket.bind(address)
 self._socket.listen(backlog)
 address = self._socket.getsockname()
- if type(address) is tuple:
- address = (socket.getfqdn(address[0]),) + address[1:]
 self._address = address
 self._family = family
 self._last_accepted = None
-- 
Amaury Forgeot d'Arc
