"Hal Murray" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED]
> What's the typical usage pattern of pool clients? (Does the > word "typical" even make sense?) > > Are they left running all the time or are they powered > off when not in use? > > If they are on all the time, schemes like replace the least-good > server every 24 hours would be a good fit. If they get powered > off then the 24 hour timer will never get reached. > > Do we need to cache the pool info over boots? If so, do we need > to occasionally reconfirm that it is still valid? It's not in itself a problem if the 24 hour timer never ticks. The client will simply resolve the pool at startup as it does now, and come up with new servers every time, distributing the load. Obviously, in that case pool info (IP addresses at least) should *not* be cached. For always-on machines, not caching results in worse time quality shortly after startup than might otherwise be achieved, but I don't see it as a big problem. If you want better time, find better servers. We _are_ talking Pool clients. Perhaps something like iburst could be implemented. Start with an interval shorter than 24 hours, back off exponentially. But maybe wait a day first, to catch those on-off systems that really don't need it. Groetjes, Maarten Wiltink _______________________________________________ questions mailing list questions@lists.ntp.org https://lists.ntp.org/mailman/listinfo/questions
