On 2013/7/2 20:46, Roy Golan wrote:
On Tue 02 Jul 2013 03:29:47 PM IDT, lofyer wrote:
> On 2013/7/2 20:15, Roy Golan wrote:
>> On Tue 02 Jul 2013 03:09:08 PM IDT, lofyer wrote:
>>> On 2013/7/2 19:58, Gianluca Cecchi wrote:
>>>> On Tue, Jul 2, 2013 at 1:17 PM, lofyer wrote:
>>>>> On 2013/7/2 19:06, Roy Golan wrote:
>>>>>> On Tue 02 Jul 2013 12:48:44 PM IDT, lofyer wrote:
>>>>>>> On 2013/7/2 17:07, Roy Golan wrote:
>>>>>>>> On Tue 02 Jul 2013 12:00:12 PM IDT, lofyer wrote:
>>>>>>>>> Will migration succeed between two different CPU?
>>>>>>>>> Suppose node A is Core-i5, node B is Xeon e5405.
>>>>> Unfortunately no, I'll paste log file later.
>>>>>
>>>>
>>>> I think you have to choose the "lowest" level of cpu type of
your
>>>> hosts for your cluster cpu type...
>>>> What is now thsi cpu type?
>>>> Gianluca
>>> "Lowest" cpu level? Is there a list of cpu priority?
>>> For now the cluster cpu type is Nehalem, there is an Core-i5 node and
>>> Xeon node in it.
>>
>> This is already enforced by the cluster so if there was a problem
>> cpu-wise you would see it.
>> Nehalem is lower then the SandyBridge (this is sorted out in the
>> drop-down)
>>
>> bottom line we need the log.
>>
> OK, I see. I would paste the log file in a few hours(the Core-i5 nodes
> has been shutdown by my friend..)
> So, the priority is AMD G5 > G4 > ... > Intel Haswell > .. Conroe
> family ?
> Unless G5 node migrate VM initiatively, it would be impossible to
> migrate VM from G5 node to others?
The cluster name (the family) sets all VM's CPUs to be virtualized as
Nehalem. The Corei5 extra instruction-set are not being used and
that's how we
ensure a migration between hosts.
Thank you all. The migration following the rule works fine.