Chuck, Thanks for the reply. I should of mentioned that I have other servers in my config.
I am a NTP Pool Member (http://www.pool.ntp.org/user/mm_202), and I checked to make sure that those S1's were 'OpenAccess' before I used them. But yes, you are right, I should limit the polling on those S1's (ntpd was just restarted). Thanks again, MM. On Tue, Jul 21, 2009 at 6:38 PM, Chuck Swiger<[email protected]> wrote: > Hi-- > > On Jul 21, 2009, at 2:30 PM, mm_202 wrote: >> >> This may be a stupid question, but when you have multiple S1's that >> differ greatly, how do you know which one to trust? > > Try adding some stratum-2 or -3 servers, and see which of the S1's tend to > keep time close to that of the other sources. > >> I have 3 configured (Im not going >> to keep all 3, but just setup the server in a new colo and want to see >> which one is >> best), and they are almost 20ms off from each other. >> >> *216.14.97.75 .CDMA. 1 u 67 128 377 32.364 1.141 >> 3.421 >> +68.216.79.113 .ACTS. 1 u 59 128 377 59.689 18.731 >> 1.403 >> +128.118.25.5 .WWV. 1 u 60 128 377 22.239 -18.805 >> 1.877 >> >> All three use different methods, are certain methods less reliable than >> others? > > The intersection algorithm chose the timesource in the middle. Most > timesources use GPS-derived clocking; but the older WWV is a shortwave radio > broadcast, and I believe ACTS is a dialup modem timesource that can also > provide a reliable PPS signal. Historically, I've found that WWV > timesources do tend to have a ~20 ms offset lagging behind the time reported > by other sources. > > Also, of course, you should not be polling a stratum-1 faster than 1024 or > maybe 512 seconds without making prior arrangements with them.... > > Regards, > -- > -Chuck > > _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
