new ovdb feature

[Heath Kehoe]

>[...]
>Now, I have had the overview database get locked up a few of the times
>these ovdb_swervers have gone down, so I've had to restart things from
>the beginning two or three times, without problems, so I guess there
>is still some risk.
>
>Maybe I will give recompiling with a lower optimization a try...
>

In my stress testing, I was using the default options.  But one thing
that is different is that I am compiling with -g, and no optimization.

Which leads me to a theory:  The loops enclosing delclient() have
"i < numclients" as the conditional.  But delclient() is decrementing
numclients (which is a global), so maybe your optimizer makes the loop
not 'see' that change.

I moved the "numclients--" out of delclient() and into the enclosing
loop.  I also rearranged things a bit so that delclient() is only
called in one place (instead of two) to simplify things a bit.

New version now in:
http://www.avalon.net/~hakehoe/inn/ovdb2/ovdb_server.c

-heath