[time-nuts] Observations of NTP during the leap second

[Martin Burnicki]

Deirdre O'Byrne wrote:
> My setup is a raspberry pi running Linux 4.1.19 and ntpd version 4.2.6p5.
> It's being fed a PPS signal from a GPS receiver, and also checks the time
> against the ntp pool.
> 
> For the second starting at 23:59:58.0 UTC, NTP was reporting second number
> 0xdc12c4fe, and had its leap second flag set.
> 
> For the second starting at 23:59:59.0 UTC, NTP was reporting second number
> 0xdc12c4ff, and the leap second flag was *not* set!
> 
> For the second starting at 23:59:60.0 UTC, NTP was again reporting second
> number 0xdc12c4ff, and the leap second flag was not set.
> 
> For the second starting at 00:00:00.0 UTC, NTP was reporting second number
> 0xdc12c500, and the leap second flag was not set.
> 
> I would have thought NTP would have the leap second flag set at least until
> 23:59:60.0.

This was a bug in in ntpd 4.2.6, which has been fixed in 4.2.8. See:
http://bugs.ntp.org/show_bug.cgi?id=2250

Martin