[ntp:questions] Very large offset and jitter values after reboot

David Woolley david at ex.djwhome.demon.co.uk.invalid
Mon Aug 25 12:13:40 UTC 2008

Nicola Berndt wrote:

> I konw, but setting up pps failed due to noise on my pps line, wich I 
> had not time to really investigate yet. There is a thread connected to 

Were you using a direct ISA or PCI interface for the PPS signal?  If not 
you can expect a large jitter, if not the multiple counts that you 
reported.  Is the PPS being properly level converted to RS 232C level?

> Thx for the iburs hint, I will try that.

It's been pointed out that iburst doesn't work for reference clocks. 
There is an implicit assumption that reference clocks are used on proper 
servers and that people only want speed ups on pure clients.

> Still very strange is that offset, wich also occurs when using 
> network-server from a ntp-pool..

I'm not sure that I would expect better than 1 second accuracy, reading 
the RTC from the kernel (this is the sort of thing that may vary between 
kernel releases, though).  You may find that using hwclock to read the 
RTC does wait for the second boundary.  (The RTC can only be read to one 
second, however, one can wait for the change in seconds.)

You should also expect a significant skew between NMEA time and the true 
seconds, which will be exacerbated by the USB interface.

More information about the questions mailing list