I believe that the /etc/ntp.conf in xntpd3-5.93-2 is incorrect. The problem is the undisciplined local clock setup: server 127.127.1.0 # local clock fudge 127.127.1.0 stratum 0 If the user runs NTP without configuration, this is OK. But if he or she sets up a few NTP servers and forgets to delete this local server, then it breaks - because this clock is at stratum 0 it *is* used preferentially. At least, it looks that way when I tested it. I'm not sure how to fix it. Leave out the local clock entirely (it's not clear it's necessary), or maybe set it at stratum 14 so it's only used as a last resort. I'm not an expert with NTP, though, so testing a fix would be a good idea. I'm a big believer in NTP. But people tend not to set it up because the NTP docs are awful. It'd be nice if the RedHat configure / install stuff made it easy for users to use NTP.
Changed ntp.conf to say fudge 127.127.1.0 stratum 10 (Fixed in dist-6.0/xntp3-5.93-6.src.rpm)
I recently did a survey of the NTP network, and found that there are at least 250 Red Hat hosts out there that still have this misconfiguration. Turns out this accounts for more than half of the bad stratum 1 clocks out there on the Internet. The bug has been fixed in new versions, so all will be better in the future, but it's interesting how long these problems persist. My discussion of this will be online at http://nelson.www.media.mit.edu/people/nelson/research/ntp-survey99/