[ntp:questions] Re: ntpdc vs iburst

Harlan Stenn stenn at maccarony.ntp.org
Fri Nov 19 08:50:31 UTC 2004


>A) The ntpd is started from ip-up (when the computer has finished the ppp
>negotiation and already connected). Plus several servers
>with iburst in ntp.conf. And ip-down stops ntpd. It works, I tested it.

However ntpd does not discipline your clock when it is not working.

>B) The ntpd starts up with the computer with only one server: 127.127.0.1.
>The ip-up adds several server with ntpdc's addserver command.
>And the ip-down removes all of them except for the local clock.
>The only one drawback of this option is that this way ntpd
>sets the clock a bit slowly.

I don't think you need the local refclock if this is the only computer
you want to sync.

And it is too bad that addserver does not support "iburst".

It might be better to have ip-up stop the running ntpd and then restart
ntpd, using several servers and iburst.  Have ip-down unpeer the servers.

This way, ntpd stays up (mostly) and should be able to keep your clock
working even when you have no connection.

I assume you are using DHCP and get a fresh address with every connection.

If not, you can leave the servers configured and adjust maxpoll (I think).

H



More information about the questions mailing list