[ntp:questions] Re: GPS Truetime NTP server and offset/jitter peak

Ralf Gross usenet at ralfgross.de
Wed Oct 19 08:30:46 UTC 2005


Martin Burnicki wrote:
> Ralf Gross wrote:
> 
>> Martin Burnicki wrote:
>>> We've read here about at least one other type of GPS receiver which was
>>> going to insert the leap second at the end of September instead of the
>>> end of December.
>>> 
>>> Do you have the possibility to log the GPS serial output across UTC
>>> midnight in order to verify that?
>> 
>> As additional info...
>> 
>> I just stopped the ntpd and used minicom to get some config values of the
>> Truetime GPS device.
>> 
>> It shows that the next leap second will occur on 12/31/2005 (exact output
>> on serial console "12/31/2005 1 13". Daylight saving is off. Timezone is
>> +00:00.
> 
> That looks good.
> 
> Unfortunately I'm not familiar with that GPS device, but just to be sure:
> Since you write you have additional time servers available, maybe you can
> stop this one and just log the serial output of the GPS receiver over
> night. 
> 
> I don't know whether the receiver can output a leap second announcement in
> its time string. But if it can, it could be helpful to know whether it
> already does the announcement or not.

I'll run the following from a cron job at 2am.

cat /dev/true0 > /tmp/true0.txt& sleep 60; kill $!

At the moment I've no better idea how to capute the output. This results
in an error message in the ntpd log file, but for 1 minute at 2am, this
shouldn't be a problem.

19 Oct 10:25:38 ntpd[32740]: clock read fd 11: Resource temporarily
unavailable

> Also the lines of your ntp.conf file which are related to the GPS receiver
> might give any clues.

server  127.127.5.0 prefer minpoll 4 maxpoll 5  # TrueTime
fudge   127.127.5.0 





More information about the questions mailing list