Re: What do people do about large clock scews
On 2009-02-22_01:46:23, Harry Putnam wrote:
> Ron Johnson <ron.l.johnson@cox.net> writes:
>
> > On 02/22/2009 12:30 AM, Harry Putnam wrote:
> >> I have debian lenny in several vmware applications on different
> >> windows machines so some of them don't get run too often.
> >>
> >> I noticed firing up one that hasn't been run for a few weeks that the
> >> time is off a by several of hours.
> >>
> >> To get it up to speed, looks like I'd have to shut ntpd down, run
> >> ntpdate then restart ntp.
> >>
> >> I'm thinking this might be a bit of a problem in other situations
> >> than my experimental setup.
> >>
> >> I'm guessing people are writing their own init script to run ntpdate
> >> on boot ahead of ntp, but I'm not yet conversant enough with the init
> >> script maze to know how to do that handily.
> >>
> >> I can script something to run ntpdate alright but getting it timed to run
> >> ahead of ntp may be a bit more daunting.
> >>
> >> But first, is there already a defacto way of doing this?
> >
> > AFAICT, ntpdate is run when networking is started, before ntp is run.
>
> Well, that should rule out a large clock skew then.
Look into the package chrony. It has a system of fine tuning the
effective rate of the hardware clock, rather than making set
corrections. It works very nicely for me. But as always, YMMV.
--
Paul E Condon
pecondon@mesanetworks.net
Reply to: