[ntp:questions] Leap Second on NTP server at stratum 2

Jochen Bern Jochen.Bern at LINworks.de
Thu Jun 11 11:36:10 UTC 2015


On 06/11/2015 08:41 AM, Kashif Mumtaz Tahir wrote:
> Dear Jochen,
> You extracted description is right , we are at stratum 2 and just syncing
> its time with stratum 1 level GPS device.
> 
> Litte bit confused with your conclusion. When leap second will happen on GPS
> what will the impact on our stratum 2 level server and below beyond ( Straum
> 3 client etc )

The part I cannot answer is whether (and when) the GPS device will
forward the information about the upcoming leap second into the data it
hands out via NTP. As an example, Meinberg states that their GPS
receivers will start announcing the leap second during the last 59 minutes:

https://www.meinbergglobal.com/english/info/leap-second.htm#refclocks

and confirms that those units which also speak NTP will include the
refclock's announcement in their replies:

https://www.meinbergglobal.com/english/info/leap-second.htm#ntp-server

So, *if* your GPS unit were a Meinberg LANTIME model, your stratum 2
server would be made aware of the upcoming leap second about one hour
before it happens.

Now, assuming that that *does* happen:

ntpd *does* forward this information (because you have no leapsecond
files overriding the info received from the server), and ntpds actually
start polling their servers more frequently as the leap second slot
draws near, so all devices running NTP (not SNTP) will typically have
their OS "armed" (informed that it - the OS, not ntpd - will have to
take action to conform to the leap second) in time.

That implies that *how* they actually do that is up to every OSes'
choices and implementation. The most usual *choice* is to have the OS
clock stepped back one second, 23:59:59.999... -> 23:59:59.000... .

And then there's the possibility that the code, which currently gets
executed once every couple *years*, has a bug making it do *something
else*. For (a historic) example, freeze the server. :-/ That's why
having test machines go through a simulated leap second is a thing.

(For sake of completeness, any client doing SNTP will notice a
one-second offset the first time it contacts its servers after the leap
second, and perform the step *then*. That should IIUC (still) include
most of the machines running Windows.)

Regards,
								J. Bern
-- 
*NEU* - NEC IT-Infrastruktur-Produkte im <http://www.linworks-shop.de/>:
Server--Storage--Virtualisierung--Management SW--Passion for Performance
Jochen Bern, Systemingenieur --- LINworks GmbH <http://www.LINworks.de/>
Postfach 100121, 64201 Darmstadt | Robert-Koch-Str. 9, 64331 Weiterstadt
PGP (1024D/4096g) FP = D18B 41B1 16C0 11BA 7F8C DCF7 E1D5 FAF4 444E 1C27
Tel. +49 6151 9067-231, Zentr. -0, Fax -299 - Amtsg. Darmstadt HRB 85202
Unternehmenssitz Weiterstadt, Geschäftsführer Metin Dogan, Oliver Michel


More information about the questions mailing list