[ntp:questions] Re: Stratum1 ntp problems

Martin Burnicki martin.burnicki at meinberg.de
Tue Aug 9 07:53:52 UTC 2005


Eugen,

Eugen COCA wrote:

> Supposing I have a Stratum 1 time server and the time delivered by the
> GPS receiver is wrong (without searching the cause for the moment). The
> result may be like this:
> 
> ntpq> host xxxxxtime.xxxx.xx
> current host set to xxxxxtime.xxxx.xx
> ntpq> pe
>      remote           refid      st t when poll reach   delay   offset
> jitter
>
==============================================================================
>  LOCAL(0)        LOCAL(0)        12 l    4   64  377    0.000    0.000
>  0.004
> *GENERIC(0)      .GPS.            0 l   62   64  377    0.000  2000.00
>  0.489
>  PPS(0)          .PPS.            0 l  711   64    0    0.000    0.000
> 4000.00
> 
> (captured Fri Aug  5 14:58:15 UTC 2005).
> 
> The server is in sync, reporting Stratum1, and the time is 2 seconds
> away.
> 
> Are there any tricks to avoid this very unhappy problem ?
> 
> P.S. The example is from a real, well known Stratum 1 time server
> (technically speaking, it's irrelevant to write the real name here).

Exactly which version of ntpd is running on that machine? There have been
some versions in the developer tree that accepted a refclock even though
the refclock claimed to be NOT synchronized. This could result in such
behaviour, that ntpd accepts the clock anyway and distributes the time of
an unsynchronized GPS receiver.

Can you verified directly at the GPS receiver that the receiver is indeed
synchronized? 

Alternatively, you could run 

ntpq -c as

to get a table of the associaton IDs, then run

ntpq -c "cv assid"

with assid being the assID value corresponding to the GENERIC clock, in your
case from the 2nd line of the billboard.

This should report what the parse driver has indeed read from the GPS
receiver and may contain information whether the receiver is synchronized
or not. If in doubt, please post the output here.


Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany




More information about the questions mailing list