Victor Duchovni:
> On Mon, Jan 17, 2011 at 08:01:18PM -0500, Wietse Venema wrote:
>
> > Perhaps postscreen should warn when the *average* delay goes over
> > some threshold. That would reduce the number of false alarms.
> > Maybe I can slip this in before the stable release.
>
> alpha smoothing should do the trick, in this case we can start with
>
> t_0 = 0,
>
> and set
>
> t_{n+1} = 0.95*t_{n} + 0.05*delta
There are many ways to arrive at a moving average. Where do these
magic numbers come from?
Wietse
> what remains to tune is how often to log warnings, the minimum
> spacing between warnings could be either based on the number of
> updates since the last warning, or elapsed time.
>
> --
> Viktor.
>
>
