[ntp:questions] basic questions about the leapsecond

Unruh unruh-spam at physics.ubc.ca
Tue Dec 16 07:32:16 UTC 2008


cmadams at hiwaay.net (Chris Adams) writes:

>Once upon a time, Unruh  <unruh-spam at physics.ubc.ca> said:
>>Ie, it would seem that if you are getting your time from the Garmin GPS, at
>>midnight on Dec 31, ntp will suddenly find that it is out by one second.
>>Since that is longer than 128ms, it will jump the time by resetting the
>>system clock by a second  the next time it queries the GPS receiver
>>input ( somewhere
>>between 16 and 1000 seconds later, depending on the poll interval for the
>>GPS receiver.)
>>
>>Ie, for a while your system time will be out by a second. 

>If you have both a NMEA GPS unit (AFAIK there's no NMEA sentence to
>alert for pending leap seconds, and even if there is, there doesn't
>appear to be support for such in ntpd) and NTP servers specified (that
>do set the leap second flag in advance), will ntpd learn about the leap
>second from the other servers and use it (even though the GPS/PPS unit
>is the "system peer")?

>From reading the ntp docs, it seems that if you download the leapseconds
file, call it /etc/ntp.leapseconds  and restart ntpd, then ntpd will read that
 file, and set the leapsecond flag on your system in the last month or day. 

I have not tested this ( my only machine running ntpd has the leapseconds
flag already set-- probably from the other two servers other than the GPS
shm refclock server which does not set the leap flag.).





More information about the questions mailing list