I wrote:

> > I'm still having problems with my serial-consoled 1.3 box not responding
> > to incoming requests. (IP filtering and NAT still works, just not
> > services on the machine itself.) I pressed Interrupt and did a ps, and
> > here's what I got (indented for reading at 80 columns):
> > -----
> [...]
> > -----
> > To my untrained and unaided eye, I see too many instances of inetd. (The
> > httpd count is a bit high too, but reasonable considering my Apache
> > setup and server load.) Weirdness in inetd would account for weirdness
> > in inetd-managed services... but what sort of weirdness is this? I don't
> > know how to track it down. The only bit of data I can determine is that
> > the problem seems to arise after about 2 days of uptime.

Ken replied:

> You're not starting apache from inetd, are you?  Especially if you have
> the option "nowait" for the httpd line in /etc/inetd.conf, it might
> possibly cause a race condition and too many running daemon processes.

A good guess, but Apache is running standalone. :(

- Amitai

-- 
Those who are too smart to engage in politics are punished by being
governed by those who are dumber.
     -- Plato