I had this problem also, never really put my finger on what was causing it,
however I have a feeling that is was due to the NTP server being a different
NTP version.
Whilst there was nothing in the logs about this, the clue that lead me to
being different versions was that:
Running ntpdc utility:
entering in the 'sysstats' command and observing the 'old version packets'
'new version packets' fields
note that each ntpdc query sends a packet of the _old_ type to the ntp
daemon. i.e. each time you sent a 'sysstats', old version packets increments.
Anyway to cut a long story short i changes the NTP server I was syncing to
and the problem went away.
currently I am syncing to 'apphys16.mst.csiro.au' which seams to play nicely
with the current debian ntp daemon;
Let me know how you go. I wouldn't mind submitting a bug report, only i
wasn't able to isolate the bug for the report to be meaningful.
Cheers
David
Sorry for the noise but it may be helpful. My current ntp.conf is:
cureton@cuscus:~$ cat /etc/ntp.conf
# /etc/ntp.conf, configuration for xntpd
# ntpd will use syslog() if logfile is not defined
logfile /var/log/ntpd
driftfile /var/lib/ntp/ntp.drift
statsdir /var/log/ntpstats/
statistics loopstats peerstats clockstats
filegen loopstats file loopstats type day enable
filegen peerstats file peerstats type day enable
filegen clockstats file clockstats type day enable
#server melbourne-fs
server apphys16.mst.csiro.au
#server tictoc.tip.csiro.au
#server ntp.cs.mu.oz.au
#server ntp.marine.csiro.au
#server ntp.mel.nml.csiro.au
#server ntp.nml.csiro.au
#server ntp.per.nml.csiro.au
#server time#server.cs.umb.edu
#server vtserf.cc.vt.edu
#server ntp-0.cso.uiuc.edu
#server ntp-1.cso.uiuc.edu
#server ntp-2.cso.unic.edu
# new #server
#server bernina.ethz.ch
#server black-ice.cc.vt.edu
#server clepsydra.dec.com
#server clock.psu.edu
#server clock.tricity.wsu.edu
#server constellation.ecn.uoknor.edu
#server delphi.cs.ucla.edu
#server esavax.esa.lanl.gov
#server fartein.ifi.uio.no
#server fuzz.psc.edu
#server gilbreth.ecn.purdue.edu
#server harbor.ecn.purdue.edu
#server heechee.esa.lanl.gov
#server jane.jpl.nasa.gov
#server kuhub.cc.ukans.edu
#server lane.cc.ukans.edu
#server libra.rice.edu
#server louie.udel.edu
#server molecule.ecn.purdue.edu
#server nic.near.net
#server noc.near.net
#server ntp.css.gov
#server ntp.univ-lyon1.fr
#server ntp0.strath.ac.uk
#server ntp1.strath.ac.uk
#server ntp2.strath.ac.uk
#server ntp3.strath.ac.uk
#server ntp4.strath.ac.uk
#server ntp0.cornell.edu
#server time.smu.edu
#server ntp-0.cso.uiuc.edu
#server ntp-1.cso.uiuc.edu
#server ntp-2.cso.uiuc.edu
#server salmon.maths.tcd.ie
#server sundial.columbia.edu
#server tick.cs.unlv.edu
#server time#server.cs.umb.edu
#server tock.cs.unlv.edu
#server vtserf.cc.vt.edu
#server wuarchive.wustl.edu
#broadcast 192.168.1.255
On Thursday 05 December 2002 16:16, Bill Moseley wrote:
> I setup a machine for a friend and every few days I ssh in to see how
> things look. Twice now I have found the date about twenty minutes behind.
> The first time I found this I ran ntpdate, made sure the hwclock was
> updated, restarted ntp-simple and thought the problem was fixed. I just
> checked again and it's twenty minutes slow again. The machine has not been
> restarted for ten days.
>
> This is on an old Dell PIII machine.
>
> What steps should I follow to reset the clock (and hwclock)? Do I need to
> remove or reset a drift file? What could cause the clock to get that far
> behind while ntpd is running?
>
> I can see that ntp is running from ps, and I see the drift file being
> written to (just a few minutes ago):
>
> $ ls -l /var/lib/ntp/ntp.drift
> -rw-r--r-- 1 root root 7 Dec 4 20:08
/var/lib/ntp/ntp.drift
>
> although /var/log/ntpstats/loopstats is old (maybe that's run once a day?)
>
> -rw-r--r-- 2 root root 1423 Dec 3 01:03 loopstats
>
> Thanks,
>
>
>
> --
> Bill Moseley
> mailto:[EMAIL PROTECTED]
>
>
> --
> To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> with a subject of "unsubscribe". Trouble? Contact
[EMAIL PROTECTED]
>
>
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
