On Sat, 29 Jan 2000, Shaul Karl wrote: > > Henrique M Holschuh wrote: > > > ntpdate is used to do a "one time only" update to your clock. ntp is used > > > to > > > discipline your clock and will in fact keep the RTC in a short leash > > > updating it every 11 minutes.
> It is my understanding that this is not true any more. It used to be true but > I believe that updating the bios clock every 11 min by the kernel is not done I just run a few tests here (2.2.14). The 2.2.14 kernel DOES have the 11 minute mode, ntp DOES enable it, and it DOES sync the RTC to system time. Sort of syncs the RTC, anyway. It looks like the kernel only syncs the seconds and a few minutes, leaving the rest of the RTC time untouched! I ran two tests, one resulted in the few seconds mismatch being fixed and perfect sync after about 20 minutes (ntp takes time to start...), the other kept the RTC in perfect sync (fixed second and minutes mismatch), but 2h30min behind the system time. Come to think of it, I recall someone was bitching about something related to the kernel, time, and 30 minute increments in the linux kernel ML, and how this is a pain for people in 15-minute timezones. My guess is that the kernel might be syncing the RTC to the second inside a 30-minute bondary, or something equally obscure. Damn, this means hwclock --systohc IS need in the *shutdown* code after all (never run it while ntp is running). hwclock --adjust should still NOT be used if you use ntp or ntpdate, though. > server. The current situation, if I get it correctly, is that > /etc/init.d/hwclock > takes care of writing the system clock to the bios just before that system > goes down. In Debian, if you have util-linux installed. Yes, you're correct. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh
