|
|
|
|
|
|
|
|
|
|
xen-devel
Re: [Xen-devel] comprehensive documentation of clock-related options for
It's also worth pointing out that NTP synchronisation in dom0 does not
affect HVM guest synchronisation at all. And in general that would be very
hard to do in a way that isn't fragile. The only solutions I can see here
for absolute time synchronisation are a PV driver or daemon for time
handling (which we do not currently have for any type of HVM guest) or
perform NTP synchronisation in the guest itself (our time handling should be
accurate enough for an ntp daemon to latch onto it).
-- Keir
On 16/02/2009 15:18, "Dan Magenheimer" <dan.magenheimer@xxxxxxxxxx> wrote:
> I agree it is confusing. There is no good default. Your
> best plan is to specify timer_mode=2 in the config file
> for hvm x86_64 linux guests.
>
> I think the defaults in XendConfig.py are to ensure that
> old saved guests don't get confused by a different default.
>
>> -----Original Message-----
>> From: Guillaume Rousse [mailto:Guillaume.Rousse@xxxxxxxx]
>> Sent: Monday, February 16, 2009 3:40 AM
>> To: xen-devel@xxxxxxxxxxxxxxxxxxx
>> Subject: [Xen-devel] comprehensive documentation of clock-related
>> options for HVM guests
>>
>>
>> Hello list.
>>
>> Despite using ntp, we're facing heavy clock skews on our HVM guests,
>> which happen to be almost all x86_64 linux. I've been
>> browsing the list
>> archive, I found reference to two different xen guest
>> configuration options:
>> - timer_mode
>> - hpet
>>
>> However, I couldn't find neither reference documentation, neither on
>> their default values.
>>
>> According to this tread, the default for timer_mode was
>> changed in some
>> part of the source tree from 0 to 1:
>> http://lists.xensource.com/archives/html/xen-devel/2008-10/msg
> 00261.html
>
> But according to this one, there was a discussion for changing
> timer_mode from 0 to 2, and hpet from 0 to 1:
> http://lists.xensource.com/archives/html/xen-devel/2008-06/msg00095.html
>
> And according to XendConfig.py code (3.3.1), the default value for both
> of them is actually 0...
>
> This is a bit confusing :(
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
|
|
|
|