David L. Mills
mills at udel.edu
Wed Feb 21 15:10:21 UTC 2007
The current implementation precludes a per-association minimum delay
calculation. The new implementation I mentioned some time back is
capable of that and more, but is stalled in the development stage.
Pending some volunteer to complete and maintain the new implementation,
the features you request will not be available.
> From looking at the code, the minimum delay (mindly) for huffpuff is
> calculated from the current sys.peer. This seems to cause me some grief
> as when ntpd changes sys.peer to a different server, with a different
> minimum delay, I get a systematic offset untill the huffpuff period is
> over and the delay from the old server ages out.
> Obviously using as short a period as possible for huffpuff will help, as
> will minimising clockhopping , but would a better soultion be to do the
> huffpuff calculation on each assciation?
> Also is there any way to get out the huffpuff values other than running
> with -d ?
More information about the questions