[LEAPSECS] The leap second, deep space and how we keep time -Brooks

[mike at lumieresimaginaire.com]

Le 28.01.2015 10:07, Poul-Henning Kamp a écrit : 

> --------
> In message <20150128021831.GA31019 at ucolick.org>, Steve Allen writes:
> On Tue 2015-01-27T21:41:17 +0000, Matsakis, Demetrios hath writ: Equally unfortunate is that 30 servers in the NTP pool inserted a leap second last Dec 31. There is no action that the ITU-R can take which will change this kind of misbehavior in these already-deployed systems.

Bullshit.

Problem:
 "On hearing certain loud noises, the patient starts shooting
 wildly in all directions."

Excellent stop-gap solution:
 "Stop making certain loud noises".

It would certainly improve things a lot, because clients, which may
upgrade at a different rate than servers, could just ignore any
server which indicates a leapsecond.

 Did one of those wild shots just get you in the foot Poul-Henning? 

I was fortunate in that I use a mixture of local and pool servers and
those pool servers that were assigned to my clients did not exhibit the
error. But as there appears to be no sanity checks on server assignment
, or if once assigned , that go 'bad', it could be the case that a pool
client got a majority of misbehaving servers. A lot of clients are never
upgraded, or significantly less so than servers. I think that Steve's
point is that bad engineering is never going to be fixed by what is
written in a recommendation. 

Mike 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist6.pair.net/pipermail/leapsecs/attachments/20150128/b2ba8e78/attachment.html>