[ntp:hackers] smearing the leap second

Harlan Stenn stenn at ntp.org
Fri Jul 10 19:37:26 UTC 2015


Mike S writes:
> On 7/9/2015 9:03 PM, Hal Murray wrote:
>> Actually, if you are going to do smearing, NTP is a fine place to do
>> it.  It lets you implement smearing without all the effort and risks
>> of updating the kernels on all[1] your client systems.
> 
> ...and THAT is exactly why some systems still have problems with leap 
> seconds, over 40 years after they first appeared - people who do things 
> incorrectly for the sake of ease and/or expedience, instead of doing 
> them correctly.

Yes.  This is reality biting.

And what is incorrect about having the time source that these devices
follow smear the leap second for them so their time remains monotonic
and follows POSIX rules about how many seconds there are in a day?

I'm more and more liking the idea that every other month or so we either
insert or delete a leap second, to make sure that the relationship
between UT and UT1 remains "in spec" *and* we have a way to make sure
that we have sufficient 'leap second activity' to make sure the code
works properly.
-- 
Harlan Stenn <stenn at ntp.org>
http://networktimefoundation.org - be a member!


More information about the hackers mailing list