Charles-François Natali <neolo...@free.fr> added the comment: I think this could be due to the multiprocessing manager's server socket backlog value, which is a little too low: by default, it's set to 5, and the tests launch up to 3 threads and 3 processes in parallel, so if we're unlucky with the scheduling, we could get some ECONNREFUSED. Unless otherwise specified, the server uses Unix domain sockets: on Linux, when the server's socket backlog is full, connect() blocks, which could explain why it doesn't happen on Linux. It would be nice to check the behavior in case of socket backlog full on affected OSes (for example OS X or FreeBSD).
Here's a run on Linux: """ $ ./python ~/test_backlog.py 0 1 2 3 4 [blocks] """ If we get ECONNREFUSED on OS X or FreeBSD, then there's a chance it's the culprit. If not, well, no idea what's going on :-) ---------- nosy: +neologix Added file: http://bugs.python.org/file24048/test_backlog.py _______________________________________ Python tracker <rep...@bugs.python.org> <http://bugs.python.org/issue13565> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
