We need to test leap smearing :)

Hal Murray halmurray at sonic.net
Thu Dec 22 04:28:49 UTC 2022

Richard Laager said:
> I was asked to enable it in Debian, but I did not.

> Note that my understanding was that "enable" meant "compile in the  support
> such that users could choose to enable it in the config" not  "enable it by
> default".

That would be my expectation.  I haven't explored the code carefully.  If 
enabled at compile time, the main flow always does the smearing calculations 
but the timer code that sets up the offset normally leaves that at 0.

docs/leapsmear.adoc is a long rambling discussion
  By Martin Burnicki
  with some edits by Harlan Stenn
as updated for NTPsec.

> I did not enable it because NTPsec's implementation of smearing  (inherited
> from NTP Classic) does not match the smearing algorithm(s)  used by the big
> players that use smearing (e.g. Google and Facebook, IIRC). 

Google says:
  We encourage anyone smearing leap seconds to use a 24-hour linear smear from 
noon to noon UTC.

There were earlier versions which did sine rather than linear.

>From our above referenced doc, you enable it with something like:
  leapsmearinterval 86400
It might be off by a factor of two, but I think it will do what everybody else 

These are my opinions.  I hate spam.

More information about the devel mailing list