Re: [fw-wiz] Acqusition of time

From: Charles W. Swiger (chuck@codefab.com)
Date: 01/29/03


From: "Charles W. Swiger" <chuck@codefab.com>
To: firewall-wizards@honor.icsalabs.com
Date: Wed Jan 29 13:54:00 2003

On Wednesday, January 29, 2003, at 06:02 PM, Luis Bruno wrote:
>> running ntpdate or some such upon system boot is a pretty good idea.
>
> Just at boot? I'd sync the machine clock to the NTP clock periodically,
> to avoid "jumps" in time; any reason NOT to do this?

Yes and no. If you want to smoothly adjust the system clock without
"jumps", you're better off running ntpd continuously than running ntpdate
via cron on some scheduled basis. ntpd will compute the system clocks'
intrinsic drift from reference time sources, and tell the kernel to skew
the time via adjtime() rather than "jump" via settimeofday().

[ Running ntpdate at boot is the best way of correcting any major errors
with the system time. Running ntpd from there on is the best way of
keeping the system time syncronized. ]

-Chuck

        Chuck Swiger | chuck@codefab.com | All your packets are belong to
us.
        
-------------+-------------------+-----------------------------------
        "The human race's favorite method for being in control of the facts
         is to ignore them." -Celia Green



Relevant Pages

  • Re: after ids
    ... Donal K. Fellows wrote: ... If you're running ntpdate, system time will always be close to real ... ntpdate) as a prerequisite for running after-using Tcl scripts? ...
    (comp.lang.tcl)
  • Re: after ids
    ... Andreas Leitgeb wrote: ... If you're running ntpdate, system time will always be close to real ... time (which I think we can assume passes uniformly for the sake of ...
    (comp.lang.tcl)

Quantcast