ntpd does not re-query servers, when a new interface appears

[Ian Smith]

On Tue, 9 Mar 2010, Jeremy Chadwick wrote:
 > On Tue, Mar 09, 2010 at 09:27:35PM +1100, Ian Smith wrote:
[..]
 > > Yes, but it looks more like name service that's not operating, ntpd 
 > > seems to be doing its best but can't resolve the hostnames?

Right smell, wrong pooch :)  Thanks for the pointer to the ntp buglist.

 > > I've always had to restart named after losing / regaining an interface, 
 > > most noticeably after a suspend/resume (eg a low battery suspend), so I 
 > > run /etc/rc.d/named restart from rc.resume.  This looks like a similar 
 > > issue perhaps, though I don't see why restarting only ntpd would fix it.
 > 
 > named is supposed to auto-probe for interfaces at a specific interval;
 > see the "interface-interval" option.  I forget what the default is,
 > but on our servers we explicitly disable it by setting it to 0.

        // We have no dynamic interfaces, so BIND shouldn't need to
        // poll for interface state {UP|DOWN}.
        // (will this fix need to reload after suspend/resume?)
        interface-interval 0;

It's rare, maybe twice a year, that this laptop cum server suspends from 
a 2hr+ lack of power - inverter failures, rewiring etc - so restarting 
named on resume makes more sense than constant iface polling 'in case'.

cheers, Ian