[ntp:questions] Re: ntp servers reporting leap second erroneously?
marc at fordson.demon.co.uk
Sat Oct 15 06:55:37 UTC 2005
On Sat, 15 Oct 2005 00:33:24 GMT, Greg Hennessy
<greg.hennessy at tantalus.no-ip.org> wrote:
>On 2005-10-14, David L. Mills <mills at udel.edu> wrote:
>> The leap warning bit is exchanged in the NTP subnet exactly following
>> announcement by NIST. The announcement can be made up to six months in
>> advance of the leap event. Every NTP client and server on this planet
>> and beyond should show the leap to be inserted.
>How does NTP tell if the leap second is to added on Dec 31, Jun 30,
>Sep 30th or Mar 31st?
It can't. NTP simply _assumes_ leap second insertions on Dec 31 or June 30.
For leap seconds scheduled for the end of any other month the whole NTP leap
second model breaks down, especially if a leap second is announced more than a
month (or three) in advance.
As a counterexample, the GPS Signal In Space (SIS) broadcasts the exact date and
time of the next leap second insertion. Unfortunately, most(?) GPS firmware
simply translates this into a "leap imminent" bit. Since the GPS ground
controllers insert the leap announcements more than 3 months in advance of the
event, this firmware behaviour isn't much use for deciding in which month a
leap will occur.
Maybe we should petition the NAVSTAR people (and operators of all time signals)
to hold off inserting leap second announcements until the beginning of the month
in which they will occur?
More information about the questions