[ntp:hackers] NTP cores in authistrustedip.

Martin Burnicki martin.burnicki at burnicki.net
Mon Apr 25 12:42:58 UTC 2016

Hal Murray wrote:
> magnus at rubidium.dyndns.org said:
>> Indeed. Especially when considering closed environments where external
>> dependencies needs to be kept to a minimum for reliability and security
>> reasons. 
> In that sort of environment, is it reasonable to import the updated leap-file 
> by sneaker-net?

The leap second file is a very safe way to provide a top level server
with leap second information plus TAI offset, but you need to take care
that every node is updated regularly.

> Does anybody in that environment need to know the TAI offset?

I think this depends on the application. If the application works based
on TAI you have to get TAI from some source since NTP only provides UTC.

Possible ways are

- using a leap second file

- using autokey

- set up a fake NTP server which spits out TAI instead of UTC, if *only*
TAI is required and the clients know they get TAI via NTP

- use PTP which is mainly based on TAI and provides UTC correction
parameters (i.e., just the other way round than NTP does)

- using a GPS receiver which can directly provide TAI, e.g. Meinberg

Which way to go depends strongly on the preconditions and requirements.


More information about the hackers mailing list