[ntp:questions] all servers rejected

Danny Mayer mayer at ntp.isc.org
Sat May 19 17:16:15 UTC 2007


Marton Fabo wrote:
>> You'll probably want to add clock=pit as a kernel command line option.
>> The VMware timekeeping whitepaper:
>>
>>  http://www.vmware.com/pdf/vmware_timekeeping.pdf
>>
>> has more details. You can also use the VMware tools to sync the guest
>> time to the host rather than running ntp. When enabled, the tools will
>> sync the guest time every 60 seconds. The VMware community forums has
>> more information that might help, too.
> 
> Thanks, your comment was extremely helpful - this is the link I was
> looking for. Unfortunately since I didn't have as obtrusive time keeping
> problems with other OSes in VMware as with this one, I haven't came to
> approach the problem from this end on my own. Thanks again.
> 
> mortee

The important thing to understand here is that there must be ONE and
ONLY ONE daemon adjusting the system clock irrespective of the number of
O/S's being hosted by VMWare. In this case it is VMWare itself that
should be doing that. Otherwise you will have different daemon's pulling
the clock in different directions and you will see problems like this.
It is one of the reasons why ntp listens on all interfaces and addresses
but it cannot do that across multiple O/S's like this.

Danny



More information about the questions mailing list