Re: RTC problem

On Mon, Jun 30, 2008 at 05:51:38PM +0200, Aragon Gouveia wrote:
> adjkerntz(8) ?

adjkerntz, as is, does this only when doing daylight switching.
This is only true for systems running their RTC on localtime.
Without starting a discussion about running your RTC on localtime
vs UTC - some RTC chips can only do UTC.

> | By Bernd Walter <ticso_at_cicely7.cicely.de>
> |                                          [ 2008-06-30 17:30 +0200 ]
> > This is not about a specific hardware - I have a general problem.
> > Normaly I run ntpdate and ntpd - ntpdate sets the time on boot and
> > then ntpd takes over to keep it in sync.
> > What recently happend was that a server with a multiple years uptime
> > rebootet because of a power failure and the local ntp-server wasn't
> > up early enough, so ntpdate didn't set the clock.
> > ntpd didn't tune the clock either, because the offset was too big.
> > I know from debugging RTC support on arm, that the RTC only gets
> > written on explizit time setting with ntpdate, date and such.
> > But since ntpd only tunes the softclock and never sets the RTC it
> > allows the RTC to run completely unsyncronized.
> > Is there a way to regulary trigger a write to the RTC without
> > disturbing ntpd, so that the offset never gets large?
> > Of course I can configure ntpd to accept a large offset, but it seems
> > wrong to me that the RTC runs unsyncronized for a large time.

-- 
B.Walter <bernd_at_bwct.de> http://www.bwct.de
Modbus/TCP Ethernet I/O Baugruppen, ARM basierte FreeBSD Rechner uvm.