Time Problem in 5.0

[Lowell Gilbert]

Kirk Strauser <kirk at strauser.com> writes:

> At 2003-04-26T02:46:16Z, Dan Nelson <dnelson at allantgroup.com> writes:
> 
> > But ntpdate does serve a useful purpose during bootup.
> 
> Dan, something I've always wondered about: if a machine runs ntpd during
> normal operation and is rebooted mainly for periodic maintenance, wouldn't
> it's hardware clock be accurate to within a few fractions of a second during
> the downtime?  I mean, if my clock it NTP-correct at noon, and I reboot the
> machine to do a `make installworld', it will be pretty close to accurate
> when I bring it back online.  Since ntpd launches instantly when not in `-q'
> mode, what's the advantage or point of running ntpdate on boot?  Why not
> just start ntpd as normal and let it smooth over the small amount of drift?

Yep.  For cases where the drift isn't small, though, it's important. 

Some examples:  machines without a working real-time clock, dual-boot
machines that have had the time adjusted incorrectly by another
operating system, machines where the real-time clock had been stepped
on in the last boot...