[K12OSN] Re: time stopped

[Michael C Wescott]

On Tue, 2005-02-15 at 09:21, Lee Harr wrote:
> ntp is not running.

wouldn't hurt, I think

> (maybe it should be... but it should not be this bad)
> 
> It is now the next day (18 hours later ...)
> 
> [root at loftern etc]# date
> Mon Feb 14 16:20:25 EST 2005
> 
> So it has counted 2 hours in the last 18.
> 
> In /var/log/messages I see this:
> 
> Feb 14 16:20:22 loftern kernel: Losing too many ticks!
> Feb 14 16:20:22 loftern kernel: TSC cannot be used as a timesource.
> Feb 14 16:20:22 loftern kernel: Possible reasons for this are:
> Feb 14 16:20:22 loftern kernel:   You're running with Speedstep,
> Feb 14 16:20:22 loftern kernel:   You don't have DMA enabled for your hard 
> disk (see hdparm),
> Feb 14 16:20:22 loftern kernel:   Incorrect TSC synchronization on an SMP 
> system (see dmesg).
> Feb 14 16:20:22 loftern kernel: Falling back to a sane timesource now.

The usual culprit in this kind of problem is something that keeps the
CPU occupied for long periods of time, keeping it from servicing timer
interrupts. As the messages suggest, Speedstep, HD not using DMA, or
improper SMP setup are possible causes.

Other possible causes are hardware problems (i.e. stuck interrupt on
some PCI card). Anything else seem broken on the machine? Any recent
configuration changes? Have you tried rebooting to see if that clears it
up?