On 2010-07-27, Thierry MARTIN <thierry-mar...@ifrance.com> wrote: > Hello, > > Thank you for this debate :-). > > Well, to my mind, it is a matter of fact that time stepping can occur > on a server having time managed with ntpd. > (My running ntp configuration leads the service to stop in such > conditions, and I have already seen that.)
Uh, manybe you should tell us what the special config is. No it should NOT step. Something is very wrong if it is stepping. > I agree, it's probably with some special configuration and under > specific circumstances. > By the way, time step is the solution that suits my need (so chrony is > probably not the answer :-)). How in the world is stepping ever the answer except on initial startup ( and chrony can be told to step then as well)? Note that chrony also converges to the right time many times faster than does ntp and keeps time 2-3 times better than does ntp (Lichvar as reported cases where chrony was 20 times better than ntp). . > > Still, the need to be notified is here (at least, for me). > > Also, as far as I could see in the source code (version ntp-4.2.2p1), > there's "only" a log message when a time step event occurs (cf > ntpdate.c:1427). > Unless someone points me to another "handler", I think I will have to > use some home-made workaround to detect this event (CLOCK_MONOTONIC vs > CLOCK_REALTIME may be my friends for this). I suspect that your time would be better spent finding out why your server is stepping. > > Once again, thank you all! > > And have a nice day. > _______________________________________________ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions