[ntp:questions] Re: More peering theory - refclocks in the mix

Joshua Coombs jcoombs at gwi.net
Thu Mar 16 21:31:39 UTC 2006


"Harlan Stenn" <stenn at ntp.isc.org> wrote in message 
news:ywn9d5gm5e3i.fsf at ntp1.isc.org...
> He's talking about using burst on a refclock.  The problem is that 
> this is a
> logical extension of the purpose of burst to a realm where it has 
> not been
> implemented.  There is no "downside" as one will cause no extra 
> traffic; the
> intente would be to "listen to more frequent answers".
>
> For some refclocks this make sense (as they may provide timestamps 
> once per
> second), and I believe Dave has done some work on handling this case 
> for
> some refclocks.
>
> H

I tried it, and I can't observe any change on the ACTS refclock.  I 
still only get one time stamp line logged in clockstats per call.

Joshua Coombs 




More information about the questions mailing list