[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Scary bugs



On Fri, Jan 28, 2000 at 09:50:59PM -0800, Nathaniel Smith wrote:
> On Fri, Jan 28, 2000 at 05:20:05PM -0800, Ryan Murray wrote:
[..]
> > 
> > I don't think that's the problem.  Setting the hardware clock again to
> > the system time isn't the problem.  The problem is, as you mention:
> > 
> > > When the system is going down for halt or reboot, hwclock is set to the
> > > value of the system clock and this *modifies /etc/adjtime*.
> > 
> > /etc/adjtime.  There is no reason to have hwclock trying to calculate
> > drift, as either NTP will be used (which is far more accurate), or the
> > system will be rebooting between Windows and Linux frequently.
> > 
> > A new user will expect their time to follow them between OS', so I think
> > instead, the support for /etc/adjtime should be removed (comment out the
> > --adjust line in hwclock).  That's what I'm doing and don't have these
> > problems.
> 
> This has nothing to do with making the time follow them between OS's --
> what /etc/adjtime does is keep track of the drift, and adjusting the
> hardware clock to counteract it.  So it's supposed to make your clock
> more accurate in both linux and windows.

Exactly. The hardware clock is a backup. It's used only to set the system
clock when the system is boot or reboot. Because the hardware clock isn't
very accurate (but predictable), /etc/adjtime allows to adjust the
hardware clock at boot-time.

By the way, the "correction" made to the init script (commenting out
the initial correction of the drift) is a mistake. Please see my
other mail.

Cheers,
-- 
Thierry LARONDE
thierry.laronde@polynum.com
website : http://www.polynum.com


Reply to: