Re: [Ntp] [[e-mail]: New Version Notification for draft-gruessing-ntp-ntpv5-requirements-02.txt]

[Miroslav Lichvar <[e-mail]>]

On Mon, Jun 14, 2021 at 05:38:26PM +0200, Dieter Sibold wrote:
> On 14 Jun 2021, at 13:04, Miroslav Lichvar wrote:
> >   I think the protocol needs to have some way to indicate that the
> >   server has leap smearing enabled. Servers implementing leap smear,
> >   but clients not knowing about it (e.g. using its own leap second
> >   source) is a major concern in some environments.
> 
> I would prefer that leap smearing is processed on the client side only. We are going to specify the version 5. This would make the specification much cleaner and would minimize the risk of misbehavior during leap second events in infrastructures where leap smearing might be needed.

>From a clock synchronization point of view it is certainly better to
perform the leap smear on clients, but I'm not sure how practical that
is in larger networks with many different client implementations. It's
easier to configure a small number of servers than all their clients.
There would need to be some way for the servers to tell their clients
they should perform a leap smear and the smear would need to be
parametrized (e.g. time of start, smear function, duration) so they
all do it in the same way.

-- 
Miroslav Lichvar