I've hosed my clock setup

[Kevin O'Gorman]

On Sat, 23 Aug 2003, Kurt Wall wrote:

> Quoth Kevin O'Gorman:
> > 
> > 
> > On Sat, 23 Aug 2003, Kurt Wall wrote:
> > 
> > > Quoth Kevin O'Gorman:
> > > > I don't know what I did the last time I went to adjust my machine's
> > > > clock, but it seems Linux no longer talks nice to the hardware clock.
> > > > Every time I boot, the clock is off by 7 hours, and for my setup
> > > > thats usually once a day (no fault of Linux, I just have to shut this
> > > > off at night).
> > > 
> > > Kevin,
> > > 
> > > Did you ever get this straightened out?
> > > 
> > > Kurt
> > > 
> > 
> > Sort of, but it's a hack.
> > 
> > I found that the /var/log/messages stuff started having two different
> > timestamps starting partway through the boot.  That was really odd.
> > 
> > Details: my RTC is set to local time because I occasionally boot to
> > Windoze.  Timestamps were all okay up to where /etc/rc.d/init.d/syslog
> > gets run and tries to do the right thing with the clock, but from
> > then on, the kernel reported times 7 hours off, presumably through
> > klogd.  Meanwhile other things continued to report the correct time,
> > presumably through syslogd.  All claimed to be PDT times.
> > 
> > My hack was to put the line
> >         /sbin/hwclock -s --localtime            # Local hack
> 
> This sets the system time from the hardware clock. If you were
> tinkering with KDE's clock setting function, undo it. I've no
> idea _how_ to undo it, of course. Here at KurtWerks, I just point
> ntpd at some public stratum 2 time servers and let ntp do the 
> grunt work. Then again, none of my machines ever boot Windows,
> so I can set my hardware clock to UTC without worry that Windows
> will helpfully reset it.

I probably was, but I don't know how to undo it either.  I cannot even
reconstruct what I did (one of the reasons I'm only lukewarm about
GUI sysadmin tools unless they do really good logging).  I may have
to live with the hack for now.

> 
> > in /etc/rc.d/init.d/syslog as the first line of the start() function.
> > I don't understand how, but that fixed it.  What's really odd is
> > that as I read things, /etc/rc.d/rc.sysinit should have already
> > executed exactly that command.
> 
> Odd.
> 
> Kurt
>