How is ntpd data used -

[Markku Kolkka]

Bob Goodwin kirjoitti viestissään (lähetysaika lauantai, 6. 
marraskuuta 2004 21:38):
> It's interesting [to me anyway] to note that the only thing
> that changed in readings taken at sleep 5 seconds and 1000
> seconds is the "jitter."

After 5 seconds ntpd has only 1 time sample from each server 
(reach=1) and the jitter value is essentially meaningless. After 
1000 seconds each server has been queried 5 times and ntpd is 
starting to get an idea about the jitter values and the quality 
of each server, and has chosen the reference server (marked by * 
in the first column). Try ntpq again after letting ntpd run for 
several hours (or days).

-- 
 Markku Kolkka
 markku.kolkka at iki.fi