[ntp:questions] Spurious positive leap second announcement?

Martin Burnicki martin.burnicki at meinberg.de
Wed Mar 7 09:01:02 UTC 2007


Hi all,

David J Taylor wrote:
> Noted this in my even log:
> 
> Detected positive leap second announcement for 2007-04-01 00:00:00 UTC

This message comes from the Windows-specific code which handles leap second
announcements received from the NTP kernel, so it will not be displayed in
non-Windows versions of ntpd.

The NTP kernel seems to have received such an announcement from one of its
upstream servers and/or refclocks. However, ntpq/ntpdc queries to those
servers mentioned in a different posting shows that none of them had a leap
second announcement at the time you sent the ntpq/ntpdc queries. So the
announcement seems to have been cancelled at the upstream server in the
meantime.

I'm not quite sure how the leap second announcement is removed in your local
ntpd. From what I've seen at a first look at the code (variable leap_next)
the announcements are or'ed from the announcements of the upstram servers
which survive the clustering. 

So if none of the upstream servers sends the announcement anymore, the local
ntpd just seems to "forget" the announcement it has seen before. However,
I'll have a closer look at this.

Please note that leap seconds have absolutely nothing to do with time zone
settings or time zone files since leap seconds are inserted into the UTC
time scale. Of course the local times just follow the leap second since
they're generally just UTC time plus an offset of some hours.

Basically, leap second might be added or subtracted at the end of any month,
but preferably this should be done at the end of June or December, and if
required at a shorter interval it should be at the end of March or
September.

I rememeber there have been invalid leap second announcements for end of
September 2005 which came from a GPS receiver (not Meinberg ;-) which "saw"
the leap second announcement from GPS for end of December, 2005, but did
not evaluate the insertion date correctly. 

So end of March is also a candidate for faulty announcements, even though
there's currently no pending leap second announcement.


Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany




More information about the questions mailing list