This started happening just recently, and I'm not quite sure what has changed. I've never consciously done anything to xinetd's configuration.

Anyway, the short version is this:
I'll be browsing through directories using Nautilus (in Gnome). At certain points, Nautilus will freeze; usually when moving between directories. The system continues to operate normally, because I can interact with a terminal and gkrellm continues to update. It's simply that Nautilus refuses to redraw its window or respond to user events.

I always run with a terminal open, and when the freeze happens, I issue a ps -aux. Every time the freeze occurs, there's an entry like so:
process id and running times have been changed to protect the innocent...

I'm no rocket scientist, but I know the process is zombied. At this point, my only recourse has been to restart xinetd with service xinetd restart. Nautilus would then go on about its business as if nothing had happened. A bit annoying, but it's happening a little more frequently these days. In fact, it happened today when I tried to login. The very first login after a cold boot froze during the loading stage for Gnome. I mean, after I entered my password, the splash screen for loading desktop components froze. I had to switch to a virtual terminal, log in, notice that xinetd had zombied, and issue a restart before I could get into Gnome.

Has anyone come across this before? Any clues on where I might start looking? Well, besides the xinetd configuration that is. Any particular setting that I should double-check?

xinetd should be making a log somewhere
mine is
/var/log/servicelog

just off hand i'm thinking try not running "fam"
i think in the xinetd configs that might be called sgi_fam

Thanks for the info. I haven't had time to look over the log yet. I'll be sure to mention if I come across what the problem is whether it's fam or some other service.

I think I figured out what the problem was, but I'm not 100% positive.

The lines in my log (/var/log/messages) were:
The above lines were repeated many, many times.

Short and sweet version: I had been screwing with getting NFS shares between two machines. In my exasperation, I started messing with my hosts.allow and hosts.deny files. I added "ALL:ALL" in hosts.deny, and manually added hosts to allow int hosts.allow. Anyway, I think the particular arrangement I had was preventing fam from connecting via localhost.

I've made changes to try and accommodate for it, but I'll be keeping an eye out just in case.

Anyone else who comes across this thread, I located a bugzilla entry for Red Hat regarding this. It has another workaround that I may have to employ if my changes don't work. Here's a link: Redhat Bugzilla Entry

Thanks for nudging me in the right direction foo.