allowing nntp access timeout

[Eric Abrahamsen <eric@ericabrahamsen.net>]

My primary select method is nntp, secondary includes an nnml server that
gets mail vs POP3. When getting new messages the nntp access stage often
hangs with the message "Checking new news…", and then stays there
indefinitely. I once let it sit there all night, just to see if it
would. I have to cancel out with "C-g" and try again.

This has to have something to do with my internet connection. I've
noticed it happens more often the first time I wake the computer or
re-connect to the internet. After that it still happens, but not so
much. Also, while I was in the US over the summer, it happened less. Now
I'm back in China, where the connection is slower and also subject to,
um, "interference", and it happens more often than not.

Is there a way to tell gnus to bail on nntp after a set number of
seconds? When the connection works it almost always works within the
first two seconds. Or, if there's a way I can get a more fine-grained
view of what's going wrong, maybe there's a way to actually "solve" the
problem? *gnus trace* doesn't tell me anything that the message buffer
doesn't.

Ideally, a timeout wouldn't trigger the server going offline, since it's
likely to work correctly the next time I check for new messages, though
I recognize the sensible thing to do would be to set it offline after a
timeout…

Any hints would be welcome!

Eric