[ntp:questions] Re: NTP and DCF77 leap logs

David J Taylor david-taylor at blueyonder.co.not-this-bit.nor-this-part.uk.invalid
Tue Jan 3 14:02:36 UTC 2006

Serge Bets wrote:
> Hint for David: Look at 4th column of loopstats file for the frequency
> correction recalculated at each poll.
> Serge.

Thanks, Serge.  Both the two client PCs below were running the ntpd 
software which inserted the leap second into Windows, and were taking the 
time both from the Internet and two local servers.  Unfortunately, I don't 
have the loopstats for the Internet-only facing PCs, which were not 
running the newer ntpd.

On client Odin, during 2005 Dec 31 the drift changed from:

-7.449 at midnight
-7.628 at 34741 seconds
-7.485 at 85768 seconds

On 2006 January 01

500.000 at 906 seconds
drifting gradually down to
480.562 at 27190 seconds

when I deleted the drift file and restarted ntpd
-120.441 at 28797 seconds
-41.424 at 46727 seconds
-9.302 at 86295 seconds

On client Stamsund, the Dec 31 drift value was around -12

On 2006 Jan 01

-11.775 at 53736 971 seconds
-12.715 at 7135 seconds
-10.968 at 86384 seconds

so that system behaved quite reasonably.  I did not have to restart ntpd 
on that system.


More information about the questions mailing list