[ntp:hackers] Re: The Server true flag

David L. Mills mills at udel.edu
Mon May 1 16:17:11 UTC 2006


Danny,

The true flag is indeed dangerous; I would just as soon remove it. I put 
it in mainly for test and evaluation and concluded it is not a good idea 
at all. It masks the underlying issue that the serial offset can (indeed 
often) disagree with the PPS offset. I see here routinely discrepancies 
of one or two milliseconds. The proper fix for this is to increase the 
minimum distance, not to diddle the selection algorithms. Let's remove 
the true flag. As for the compatibility issue with ntpdc, you know my 
position on that.

Dave

Danny Mayer wrote:

> Dave,
>
> Can you expand the explanation of the true flag on a server
> configuration line. The confopt.html documentation doesn't mention it on
> the server command line options (an error?) but there is a paragraph
> documenting it.
>
> The reason that this came up at all is that the ntpdc code does not
> support this option and ntp_intres uses the mode 7 packets to set up
> servers once it has obtained the IP address. Personally I think this is
> a dangerous flag but there is a reason to support it. It we need to keep
> it we may need to update the ntp_config code to support it.
>
> What do you think?
>
> Danny




More information about the hackers mailing list