[ntp:questions] Re: What went wrong with the leap second
martin.burnicki at meinberg.de
Wed Jan 4 16:55:11 UTC 2006
Rob van der Putten wrote:
> Hi there
> Martin Burnicki wrote:
>> This indicates that either the NTP daemon itself, or its upstream
>> server(s), may not have received the leap second announcement, and so
>> time synchronization has been messed up.
>> Please have also a look at my posting from Jan 3 in the thread with the
>> subject "Drift handling...".
>> I don't know exacly what you mean here.
> The idea is to mask the leap second;
> Instead of adding a second you just make the seconds longer.
Since leap seconds have not been inserted at regular intervals, this would
mean that you'd have to adapt the length of the second to the earth
rotation speed, which results in seconds of variable length.
AFAIK this had been the case before atomic time had been introduced.
>> The GPS satellites start broadcasting the upcoming leap second shortly
>> after this has been announced by the IERS. The last leap second has been
>> announced by the GPS satellites since middle of July, 2005.
> I only saw them on the 31st.
Yes, but I guess you just saw what came out of the GPS receiver, or even
ntpd. That's what I had meant with the text below:
>> How these announcements are passed to ntpd depends on how the leap second
>> announcement is handled by the firmware of the DCF77 or GPS receiver, how
>> the driver module for that particular device evaluates that announcement
>> and passes it to the ntpd loop filter, and of course by the version of
>> ntpd, since there have been some modifications across the versions.
>> Maybe our leap second info page is helpful. Please have a look at
>> Hope this helps.
> Not much.
Sorry for that.
> Thanks anyway.
> The problem was that I configured 4 of my ISP NTP servers as sync source.
> They don't use their own GPS / DCF77 / whatever receivers but use several
> other NTPDs to sync to.
> I think some of their time sources broke and polluted their own time and
> that of their customers.
> I was hoping that one could piece together exactly what has gone wrong.
> Right now I don't think that this is ever going to happen.
I thought what I had posted under "Drift handling.." would have made clearer
what has happened. Some of the upstream server did the leap, others did
not, and the clients synchronized more or less randomly to either type,
applying wrong corrections and messing up their own calculated drift.
> The only way that one can avoid these problemes is probably to use ones
> own GPS receiver instead of somebody elses NTPD.
More information about the questions