[ntp:questions] Query about NTP accuracy

Richard B. Gilbert rgilbert88 at comcast.net
Mon May 25 23:39:06 UTC 2009


Harlan Stenn wrote:
>>>> In article <gvcrb4$2ltn$3 at grapevine.csail.mit.edu>, wollman at bimajority.org (Garrett Wollman) writes:
> Garrett> What is actually needed at boot time is usually neither of these
> Garrett> things.  It's more like "if the clock needs to be stepped
> Garrett> backwards, figure this out as quickly as possible, before any
> Garrett> important services start, accepting that there is some small chance
> Garrett> that an additional step in either direction may be required."
> 
> Sounds reasonable - how do we do it?

If you are going to step the clock, doesn't it make sense to step it the 
  proper amount the first time?  Especially if you are going to step it 
backward?




More information about the questions mailing list