On 18:08 06 Apr 2003, Jeremy Portzer <jeremyp@xxxxxxxxx> wrote: | On Sun, 2003-04-06 at 12:53, D. D. Brierton wrote: | > It seems like NTPD is attempting to correct an 8 second drift in my | > clock but is failing to do so, and indeed my clock is eight seconds out. | > Where can I find further details about why the "synchronisation" is | > being "lost"? Does NTPD write to a log anywhere? Why would its attempt | > to correct the clock fail? Currently it is attempting to synchronise | > with clock2.redhat.com. | | I don't know if this is the exact reason for those error messages, but | remember that ntpd has a hard time correcting large clock drifts, and 8 | seconds is large in ntpd's world. So try stopping ntpd, and then | force-resetting the clock with ntpdate (or even rdate). Then, restart | it and see if ntpd can correct the much smaller difference that will now | be present. More to the point, doing large (anything over 1 second) clock changes can confused things that pay attention to the time (eg cron or make, by no means the only examples). So ntpd _refuses_ to magic adjustments beyond a certain magnitude. You must do those by hand, just as Jeremy describes above. It's enough to issue /etc/init.d/ntpd restart because the first thing an ntpd start does is run ntpdate. This doesn't explain why you lost this time, but if you lost contact with your time servers and had sufficient drift in your hardware clock (all clocks have some drift), this can certainly happen. -- Cameron Simpson, DoD#743 cs@xxxxxxxxxx http://www.zip.com.au/~cs/ NOTWORK: n. A network when it is acting flaky. Origin (?) IBM. - Hackers' Dictionary
