[time-nuts] Legal Time dissemination
David J Taylor
david-taylor at blueyonder.co.uk
Wed Jun 5 05:58:02 UTC 2013
Hi Chris!
I have a week's worth of data now... Take a look at these graphics
http://www.flickr.com/photos/miguelbarbosagoncalves/8955346508/ and
http://www.flickr.com/photos/miguelbarbosagoncalves/8955346522/
One of them shows is the plot of the loopstats of my stratum 1 server. The
other one shows the offset determined every minute of their servers
compared to mine.
On 28 May 2013 16:11, Chris Albertson <albertson.chris at gmail.com> wrote:
> The "stratum 2" servers are by definition not connected to GPS. They get
> their time from some other NTP server that is connected to an
> authoritative
> clock which may or may not be GPS. It looks like the "red server" has
>
I know that but I would expect good stratum 2 servers if they are keeping
their stratum 1 servers private and not available to the public.
> rather smooth swings over around a "handful" of milliseconds. This is to
> be excepted. It is within the normal range of what NTP does. Perhaps it
> is a Windows PC running in some room where the temperature changes and the
> network that connects it to the strum 1 server is loaded. I don't know
> but
> a handful of milliseconds is in the normal range. Yes it could be better.
>
I don't believe it is a Windows server but I could be wrong...
> Also the curve is somewhat smooth. It does not look like noise from a
> busy
> network. It looks like their server really is moving around.
>
I believe they are on the same network. It looks like a temperature problem
really.
Thanks for all the input!
Kind regards,
Miguel
==========================================
Miguel,
I notice a step in the blue (ntp02) graph in:
http://www.flickr.com/photos/miguelbarbosagoncalves/8955346508/
and I think in the red (ntp04) graph as well. This suggests some change in
connectivity between your monitoring PC and both servers, so perhaps a
network change. Unless your stratum-1 server took a 2 millisecond step,
which seems unlikely.
Their ntp02looks well behaved, but their ntp04 is clearly not well behaved,
and disappointing for public service. In my own plotting program:
http://www.satsignal.eu/software/net.htm#NTPplotter
I find it useful to be able to plot the offset versus time of day, as
regular temperature variations tend to stand out - at least in my
environment where the heating switches on at the same time every morning.
For comparison on a similar time scale, you might like to look at my Windows
LAN-synced PCs here:
http://www.satsignal.eu/mrtg/performance_ntp.php?period=week#windows
so even if their servers were running Windows, they could be doing far
better than that your graph suggests. Could you plot offset versus time of
day?
Cheers,
David
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor at blueyonder.co.uk
More information about the Time-nuts_lists.febo.com
mailing list