Henrique de Moraes Holschuh wrote: > Pierre Frenkiel wrote: > > Anyway, the fact that this problem appeared just a few days ago on a > > machine running since about 5 years seems indicate a hardware > > problem (battery?) > > Yes, the lower "voltage" caused by a dying battery can increase the > systematic drift on the RTC, or even stop the clock entirely for a small > while. It depends upon the motherboard but most won't draw battery power unless the computer is not powered. The battery is only used when the mains power is not connected. If the mains power is connected then the system uses mains power and not the battery power. If a computer is left sitting off for a long time (years) then the battery will most likely be drained flat. But one running under power doesn't use the battery and shouldn't need it. Again this is all motherboard dependent. They don't have to do it the right way. But this is the way we did it. In any case the motherboard battery and hardware clock is only for getting time close at system boot time. After the system is booted then the OS will keep its own time based upon the system frequency which is unrelated to the motherboard hardware clock. So you basically always have two clocks. (Notably the Raspberry Pi does not have a hardware clock at all. It only has the OS system time. Only one clock there. And it works just fine that way.) Utilities such as NTP are really good at adjusting the counting per second to tune the OS view of time to be very accurate. If that isn't working then I suspect some other problem. Such as two daemons fighting each other both trying to adjust the clock. Or some other failure. Bob
Attachment:
signature.asc
Description: Digital signature