[Libvirt-cim] [PATCH] [TEST] Adding 05_RAPF_err.py to verify RAPF
Deepti B Kalakeri
deeptik at linux.vnet.ibm.com
Thu Apr 10 13:46:04 UTC 2008
Kaitlin Rupert wrote:
>>>> Ok , will modify this.
>>>>> DK> + if virt != 'XenFV':
>>>>> DK> + bridge = vsxml.set_vbridge(server)
>>>>>
>>>>> Can you explain why this special case exists?
>>>>>
>>>>>
>>>> The network information for the XenFV is getting assigned by default.
>>>> The def _devices(): section of the XenFV is calling the
>>>>
>>>> self.set_bridge(CIM_IP) and hence the bridge information is getting
>>>> assigned by default.
>>>> While this is not the case with the Xen and KVM.
>>>
>>> KVM also calls set_bridge() in the _devices() call.
>>>
>>> Xen does not, but the _devices() call for Xen sets the interface as
>>> an ethernet type. So there's no need to set a bridge for the Xen case.
>> Wrt to the tc when working with NetRASD requires the domain to be
>> created with interface type as bridge.
>> NetRASD lists the domains which are created with bridge as the
>> network type. Please correct me if I am wrong.
>
> You'll only get a NetRASD instance if the guest has an interface that
> is tied to a virtual network. That is - if the guest's bridge belongs
> to a defined virtual network.
>
>> I suggest we keep the default network type for all the three Virtual
>> types to be bridge and the give
>> another function which allows to add network type other than bridge
>> to the domain when required
>>
>
> The original Xen XML used an ethernet style network as the default.
> Can you verify which (if any) test cases will be affected if we change
> Xen to use a bridge style network as the default?
>
> In general, I'd like to see the default XML for all three types be as
> similar as possible (where it makes sense). So I think this is a
> sound idea.
>
I need to verify this.
More information about the Libvirt-cim
mailing list