"Andrew D Dixon" <[EMAIL PROTECTED]> writes: > David Schleef wrote: > > > On Fri, May 25, 2001 at 03:44:37PM +0200, Michel D�nzer wrote: > > > Russell Hires wrote: > > > > > > > I'm playing with the date command, trying to understand why it doesn't > > > > pick up the current time/date from ntpdate/ntpd... > > > > > > ntpdate sometimes fails when the date is too far off. Setting the date > > > manually and then running ntpdate usually works for me. > > > > I've noticed this as well, after I crash the machine really > > hard. The RTC gets set to 1904, and ntpdate doesn't seem to handle > > times that are outside the current epoch. Does anyone know if this > > is this a limitation or bug with ntpdate or the kernel interface for the > > real-time clock? > > I just used ntpdate to set the clock to the curent UTC from 0 (jan 1 1970) > with > out a problem so this problem probably isn't simply related to how far your > deviating from the time given by the server.
Right. I just set my clock to Jan 1 1904, and used ntpdate. amadeus:/dev# date -s "Jan 1 12:05:00 1904" Fri Jan 1 12:05:00 MST 1904 amadeus:/dev# ntpdate ntp.cs.unm.edu 7 May 07:53:19 ntpdate[16646]: step time server 64.106.20.32 offset 926106473.917452 sec amadeus:/dev# date Sun May 7 07:53:22 MST 1933 Using date to set it to Jan 1 2001, and then using ntpdate worked fine. It seems ntpdate gets confused when the date is negative compared to the epoch. jas.
