On Thu, Oct 25, 2007 at 05:16:58PM +0200, Adrian von Bidder wrote: > On Thursday 25 October 2007 16.27:16 Sam Mason wrote: > > On Thu, Oct 25, 2007 at 04:04:11PM +0200, Adrian von Bidder wrote: > > > (Hmmm.... this gets me thinking: a way to signal to ntpd that it should > > > initiate an ntp volley with all its servers right *now* would be good > > > for systems with intermittent internet connection, especially now that > > > ntpd rescans interfaces and doesn't need to be restarted if net goes > > > down and up. Combined with really high minpoll values this could be an > > > interesting thing.) > > > > Why not just use ntp-date here? > > Because ntpd corrects for clock drift, which ntpdate cannot do. AFAIK the > algorithms can deal with irregular intervals between updates from the > server, after all it changes the poll interval itself. (But I've not > really looked at the algorithms, this is just from a very shallow > understanding of how ntpd works.)
OK, good point. Maybe it would be able to extract some information out of it. If you can basically get your clock OK before your connection goes intermittent then it may have a chance. I don't know the algorithm well enough to say any more though. Sam _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
