Mailing List Archive

Question
I have a Python manager and clients. I added a feature that if the client or server are killed, they will be restarted.

When I restart the server after being killed the clients get a broken pipe error, but the connect call that worked on the first connect, doesn't work now.

When a client is restarted it runs the same connection code as always. The connect seems to complete, but the calls don't complete. They just hang.

Is there an issue with broken connections and multiprocessing managers?

I have tried this with the 3.9.0rc2 and the problem still exists.

Also, I have a single connection to a manager. I get two proxies to that manager on the one connection. The documentation says that proxies to a manager are not thread safe. What about two proxies. Do I need to use a single mutex for both proxies, or can I have one mutex for each proxy? If I have a blocking call on one proxy, would using the other proxy step on the toes of the first?


Michael L. Boom
858-240-6059
mikeboom@mikeboom.com<mailto:mikeboom@mikeboom.com>
11532 Hadar Dr.
San Diego, CA. 92126

--
https://mail.python.org/mailman/listinfo/python-list