[ntp:questions] Re: Can NTP affect the time even when not running? (with platform details this time)

mike michael.no.spam.cook at wanadoo.fr
Wed Nov 24 14:02:56 UTC 2004


Digging a bit more I see that the freq and tick kernel variables can
be set using adjtimex.  If your system is keeping the same bad behaviour 
after a boot it is probably because of the existance of a bad 
/etc/adjtime file ( mine seems to survive boot)
[mike at quark linux-2.6.8.1]$ who -b
          system boot  Nov 24 10:09
[mike at quark linux-2.6.8.1]$ ls -l /etc/adjtime
-rw-r--r--  1 root root 44 Nov 24 08:36 /etc/adjtime

Get rid of it, and go read the man pages  adjtimex and hwclock.
Yuo should be able to get back to correct values without a boot.

If you read French there is even a web page which outlines the strategy.
Probably in english also somewhere.
http://www.linux-france.org/article/sys/heure/ch2.html#s-bref


Parker Jones wrote:
> Dear all,
> 
> My hardware clock used to lose about 1 minute a week so I decided to 
> start using ntp.
> 
> I had problems with ntp that I haven't been able to resolve, so I have 
> reverted back to just the hardware clock.
> 
> But the clock now drifts (loses) several hours each day.   Is it 
> possible that ntp could have caused this? e.g.  a drift parameter set by 
> ntp?
> 
> Any suggestions much appreciated,
> 
> PJ
> Platform: linux Fedora Core 2, kernel 2.6.8-1.521, ntp-4.2.0-7 on NEC 
> Versa LX 400MHz laptop
> Will post config files and logs if it helps.
> 
> _________________________________________________________________
> Need more speed? Get Xtra JetStream  @ http://xtra.co.nz/jetstream
> 



More information about the questions mailing list