[ntp:questions] Re: maxpoll >=14 causes other clients to reject server
David L. Mills
mills at udel.edu
Mon Mar 13 02:43:18 UTC 2006
Two issues: Poll intervals longer than 10 (1024 s) work only with NTPv4
clients. Second, the firewall clients are seeing the upstream
dispersion, and thus the synchronization distance, inherited from the
firewall. If they ratchet up the poll interval, the total distance might
exceed the distance threshold, be default 1 s. You can tinker that if
needed, but I do suggest the default maxpoll of 10 is appropriate and
nonintrusive for mice. It's the elephants that poll the national time
servers at one-second intervals that are the problem.
David D. Hagood wrote:
> I have an NTP server at my firewall, synced to various sources on the
> Internet. Being a good netizen I want to poll those servers no more than
> needed, so I set the maxpoll high so that once my server is synced, it
> should poll its sources very infrequently.
> However, if I set maxpoll to 14 or more, then my computers on the inside
> will occasionally reject the firewall server as not being synced because
> it has not synced recently enough.
> I am assuming the firewall server is reporting that it hasn't synced
> recently enough, and that is why the clients are rejecting it. Fair
> enough, but then shouldn't the server go sync again, even though it has
> not reached the maxpoll interval? Alternatively, should not the server
> just say "I'm OK, I have synced within my programmed time, all is well"?
> Or lastly, shouldn't the server say "maxpoll >=14 is not valid"?
> All machines in this discussion are Fedora Core machines running
> questions mailing list
> questions at lists.ntp.isc.org
More information about the questions