[ntp:questions] Re: HOWTO prepare ntpd to the leap of a second

Serge Bets serge.bets at NOSPAM.laposte.invalid
Fri Dec 30 09:15:00 UTC 2005

Hello Richard,

 On Thursday, December 29, 2005 at 18:58:05 -0500, Richard B. Gilbert wrote:

> Does ntpd maintain a copy of this leap second file somewhere? Where?

The operator puts the leapseconds file in the keysdir. The daemon reads
the file, but does not save it, nor update it in any way. The daemon
propagates a volatile in-memory copy of the data table extracted from
the file to its clients using Autokey authentication protocol.

> I'd always assumed that NIST servers set the leap second bits when it
> was time to set them; that they propagated downward and the leap
> second happened automagically, everywhere.

Yes: That's the main scheme. Just note that not everybody is downstream
of NIST, or has a leap bits clean path from NIST. And any other server
can assert leap bits.

The optional NIST leapseconds file is complementary. The data table
gives to the NTP daemon and to its Autokey clients more infos than
simple tristate leap bits: Epoch of next leap second (roughly 5 monthes
in advance), and current TAI offset. And it gives them enough authority
to assert leap=01 during the last 30 days.

Finally clients not using Autokey and not having themselves the NIST
file are informed only by received leap bits.

Serge point Bets arobase laposte point net

More information about the questions mailing list