[ntp:questions] Re: Correcting my time servers clock drift on Alpha ES40s / Tru64

Tom Smith smith at cag.zko.hp.com
Fri Oct 21 03:31:02 UTC 2005


brett.sander at tenix.com wrote:
> Dave,
> 
> The clocks on the 2 servers (as far as I can tell) are running at the
> same time.  I will have to take a look into the orphan mode. If the
> clock of my alpha box that I designate as the server is not great at
> keeping time though, doesn't that mean my whole network will be out as
> it is now?
> 
> Below are some extracts of the ntp logs:
> 
> terrance:
> 
> 20 Jul 17:40:44 xntpd[764]: logging to file /var/adm/ntpd.log
> 20 Jul 17:40:44 xntpd[764]: ntpd 4.0.98a Wed Oct 16 17:36:24 EDT 2002
> 20 Jul 17:40:44 xntpd[764]: precision = 976 usec
> 20 Jul 17:40:44 xntpd[764]: kern_enable is 1
> 20 Jul 17:40:44 xntpd[764]: using kernel phase-lock loop 0040
> 20 Jul 17:40:44 xntpd[764]: frequency initialized -5.490 from
> /cluster/members/{memb}/etc/ntp.drift
> 20 Jul 17:40:44 xntpd[764]: using kernel phase-lock loop 0041 
> 
> phillip:
> 
> 20 Jul 17:41:50 xntpd[764]: logging to file /var/adm/ntpd.log
> 20 Jul 17:41:50 xntpd[764]: ntpd 4.0.98a Wed Oct 16 17:36:24 EDT 2002
> 20 Jul 17:41:50 xntpd[764]: precision = 976 usec
> 20 Jul 17:41:50 xntpd[764]: kern_enable is 1
> 20 Jul 17:41:50 xntpd[764]: using kernel phase-lock loop 0040
> 20 Jul 17:41:50 xntpd[764]: frequency initialized -10.384 from
> /cluster/members/{memb}/etc/ntp.drift
> 20 Jul 17:41:50 xntpd[764]: using kernel phase-lock loop 0041
> 20 Jul 18:04:47 xntpd[764]: time reset 0.556995 s
> 20 Jul 18:04:47 xntpd[764]: synchronisation lost
> 
> Is the synchronisation lost message a normal thing?
> 

It's a direct result of the reset that immediately preceeds it,
which is, in turn, a direct result of having booted (or restarted
NTP on) both systems at the same time and having no fixed time
reference other than the systems' own local clocks. phillip came
up with a time 0.55... seconds behind the time of terrance. phillip
chose to follow and once it had stepped it was unsynchronized until
it had again gathered enough samples to synchronize again.

If you have no reference to "true" time, you can have no expectation
of keeping it.

-Tom




More information about the questions mailing list