In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (Danny Mayer) wrote:

> This is even worse than having no time at all. The upstream servers are
> only synchronized to themselves, ie not at all, which is what those

No they are not.  The problem is that they are synchronised using protocols
which only have a precision of 1 second and which have no root delay 
compensation, have long poll delays and have no frequency compensation,
so have high dispersions.  I'm not sure if SMB  has 1 second precision, in
which case, with daytime, they would have two precision 1 second hops in
tandem.

> refids are telling you. They have no reference which will allow them to

The refids only tell you the boundary of the NTP part of the synchronisation
network.

> stay in synch, even loosely. You are better of synchronizing to just one
> of these, but even better to some real server that are providing proper
> time.

He is synchronizing to a real server, but using one daytime hop and one
SMB hop.  I did note that it is an NTP architecture violation to have
non-NTP hops in the path to the real reference clocks.  localclock is 
being used here in the synchronised by other means context, not in the
completely undisciplined context.

_______________________________________________
questions mailing list
[email protected]
https://lists.ntp.isc.org/mailman/listinfo/questions

Reply via email to