[pkg-ntp-maintainers] Bug#484974: Bug#484974: Bug#484974: Bug#484974: ntpdate's ip-up script runs even for virtual interfaces

[Josip Rodin]

On Mon, Jul 14, 2008 at 08:58:32AM -0400, Steve Kostecke wrote:
> >I answered this earlier, but here goes one more time: I use ntpdate for its
> >exact simplest purpose - setting the clock ad hoc from a specified NTP
> >server. The machine doesn't have ntpd installed, and I either don't want
> >it at all, or I don't want ntpd but openntpd on it.
> 
> ntpdate is not that much different from rdate (except for the fact that
> ntpdate can slew the clock and, of course, the protocol being used).
> 
> ntpdate just polls the given servers and uses the first response.
> ntpdate does not compare the responses from multiple sources to
> ascertain the correct time; so it is possible for ntpdate to use an
> incorrect time.

Well, yes, but rdate uses a different protocol, and one that tends to be
disabled everywhere these days.

> ntpd -gq is the recommended replacement for ntpdate. Unless you need to
> use ntpdate's debug/query or alternate source port features.
> 
> ntpq -gq uses all ntpd features (including Autokey/SymmetricKey
> authentication, clock selection algorithms), polls all time sources with
> iburst, and exits after stepping/slewing the clock.

Right, right, but I don't have that unless I install ntpd. And I don't want
that either because I don't want to go through the trouble of running
another daemon or disabling it, or because the machine has openntpd. :)

> >then I can only suggest that nobody's forcing you to do it... also,
> >nobody's forcing you to make random changes which "seem like a useful
> >generalization", but aren't necessarily...
> 
> I don't understand this.

Peter said earlier that the reason why he made ntpdate start in
network/if-up.d was a useful generalization of what people were asking for
(for it to be run in ppp/ip-up.d/).

-- 
     2. That which causes joy or happiness.