Le lundi 18 avril 2005 à 23:21 +0200, Richard Fish a écrit :
> Frédéric Grosshans wrote:
>
> >I'll choose the simpler cronjob solution, thanks. Anyway, isn't that a
> >kernel bug ? How sould I report this ?
> >
> >
>
> Well, from googling around, it seems that a few seconds per day of drift
> between the system clock and hardware clock is considered perfectly
> normal,
A few second/day, yes, but I don't think 5 minutes/day should be
considered as normal. (that's 1 second every 5 minutes !) Any idea where
it could come from ?
The cron solution solves the problem, but I do not like the idea of
having a system with such an intrinsic drift ! It's the only linux
desktop in the lab, and the only computer with such a drift. Not a good
point if I want to convert the others to linux ;-(
> and the adjtimex/hwclock/cron solution is really the only option
> for synchronizing the two, while NTP is the best solution. Indeed the
> normal case is for both system and hardware clocks to drift away from
> the real time.
OK
>
> You mentioned you were behind a firewall and couldn't use NTP. I am
> curious, because most networks I have connected to have only disabled
> NTP access if they provide an internal NTP server. Have you asked the
> network admins if there is an NTP server available?
To be more precise, I'm in a University which has some metrology lab,
and yes there are local ntp servers, but I don't seem to be able to
access them neither. That's probably because I'm not in any DNS. The
solution to change this would probably to be in charge of the DNS of my
lab, but :
- I'm on a temporary position here
- I'm here as a scientist, and I don't really want to become
the lab's sysadmin
Thanks for the suggestion anyway,
Fred
--
Frédéric Grosshans <[EMAIL PROTECTED]>
--
gentoo-user@gentoo.org mailing list
