[ntp:questions] Re: synchronisation lost problem

mike michael.no.spam.cook at wanadoo.fr
Sun Dec 12 16:02:40 UTC 2004


Did you find the cause of the problem?? I did not see any resolution in 
the thread.

Your problem is almost certainly not with ntp, but the clock disciplin.
ntpd seem to be reporting varied drift rate changes. There is an additional
        1.812 ppm in the first hour after the first reset
        0.86  ppm in the second
       38.776 ppm in the third

   Those are widely differing for a machine in a stable temperature 
environement as you indicate. My machine which is an unregulated 
environement doesn't change more than 0.2ppm in the same period.
  So either your hardware clock is broken, or, if you have already 
changed it (and it was not DOA) then the /etc/adjtime value is
reporting data from the previous broken hardware which will induce error 
in the clock adjustments.
  Try removing it and rebooting the machine.
  If that has already been done, try graphing the clock drift to see if 
there is a pattern you can relate to other events. It would probably be
a good idea to to graph the drift while not syncing with external 
machines (or at least measuring drift with ntp stopped for a while) to
eliminate ntp as a source.
   Also check that the selected ntp server is not flip flopping between 
the two you have.  Having just two defined servers is the worst possible 
configuration.



Serge Tchesmeli wrote:
< snip>

>  6 Dec 13:56:47 ntpd[22002]: time reset 0.142454 s
>  6 Dec 14:38:22 ntpd[22002]: offset 0.022124 sec freq 178.812 ppm error 
> 0.027763 poll 9
>  6 Dec 15:38:23 ntpd[22002]: offset 0.017784 sec freq 180.624 ppm error 
> 0.028154 poll 10
>  6 Dec 16:38:23 ntpd[22002]: offset 0.113686 sec freq 181.484 ppm error 
> 0.046980 poll 10
>  6 Dec 17:11:15 ntpd[22002]: time reset 0.184771 s
>  6 Dec 17:11:15 ntpd[22002]: synchronisation lost

>  6 Dec 17:38:24 ntpd[22002]: offset 0.051767 sec freq 220.260 ppm error 
> 0.040462 poll 8
> 



More information about the questions mailing list