Darrin Chandler wrote: > On Sat, Mar 01, 2008 at 06:38:26AM -0000, David J Taylor wrote: > >> The 20ms steps on .114 almost look like lost interrupts, or that the >> server is running Windows. Is it? >> > > These 20ms steps are bothering me. I've got two different machines > architectures running different ntp software, and these 20ms steps are > still in sync between the two servers. I've been monitoring both from > home DSL through ntpq -c peers and I'm only seeing very minor > fluctuations in offset. > > So, I'm tempted to think this is due to differences in routing or > whatever from the monitor in Europe to my boxen in the US. The machines > are in a datacenter with connections to 3 backbone providers, so it's > not unusual to see routes change periodically. Is this the likely cause? > Any other ideas? > > Waht does loopstats and peerstats on the ntpd box say.
Enable them with: statsdir /var/log/ntpstats/ # directory for statistics files filegen peerstats file peerstats type day enable filegen loopstats file loopstats type day enable filegen clockstats file clockstats type day enable You should be able to easily see if there are any gross changes and where they are coming from. John _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
