From: Dave Thomas <Dave@...>
Date: 2001-11-27T22:34:03+09:00
Subject: [ruby-talk:26652] Re: TCP hang problem

"HarryO" <harryo@zipworld.com.au> writes:

> The strange thing is that this code is multi-threaded and when this
> happens, I can't even get another request into it.  Since the accepts are
> happening in the main thread, I would have thought that a hang in the
> write wouldn't affect that.

Sounds like it might be DNS related: a DNS lookup hangs all threads.

Depending on your application, setting

          Socket.do_not_reverse_lookup = true

might help. Otherwise, you might be able to get in to the hung process
with gdb and do a backtrace to find out what's going on.


Regards


Dave