[LEAPSECS] Leap seconds ain't broken, but most implementations are broken

Ask Bjørn Hansen ask at develooper.com
Tue Jan 3 22:43:38 EST 2017


> On Jan 3, 2017, at 9:56 AM, Steve Allen <sla at ucolick.org> wrote:
> 
> Also not completely surprising was that we saw one of the pool.ntp.org <http://pool.ntp.org/>
> servers (not Google) in use by our machines was clearly making a leap
> smear available.

You were pretty lucky then. It was only about 9 servers out of more than 3,000 that made that error[1]. :-/

Many hundreds just messed it up though. 4 applied it twice.

As Tony mentioned, even national time labs messed it up.

This is why some of us roll our eyes at the suggestion that the implementations will get fixed / catch up. Rounded down there are zero signs of that happening. Leap seconds will never work properly or be implemented consistently. They just make time keeping worse.


Ask

[1] As much as I dislike the leap seconds; smearing is only appropriate if you specifically have chosen it so they’re not supposed to be in the pool.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist6.pair.net/pipermail/leapsecs/attachments/20170103/50fc3711/attachment.html>


More information about the LEAPSECS mailing list