On 7/13/20 1:50 PM, Bastien Orivel wrote:
On 13/07/2020 13:41, Michal Privoznik wrote:
> On 7/13/20 11:44 AM, Bastien Orivel wrote:
>> This is only used in the ESX driver where, when set to "no", it will
>> ignore all the checks libvirt does about the origin of the MAC address
>> (whether or not it's in a VMWare OUI) and forward the original one to
>> the ESX server telling it not to check it either.
>>
>> This allows keeping a deterministic MAC address which can be useful for
>> licensed software which might dislike changes.
>>
>
> While you get bonus points for remembering to document this change, it
> should go into a separate patch, because keeping it in a single one
> usually leads to conflicts on backports.
Oops, didn't know about that, will split in a v2 once we resolved the
other question.
> But anyway, looking at virVMXFormatEthernet() - why don't we set all
> MACs 'static'? Alternatively, we can use @mac_generated member to
> determine whether the MAC address was provided by user or
> automagically generated (and use static/generated addressType
> accrodingly)?
Mostly because I didn't want to break any existing script/setup that
relies on the fact that libvirt would set the address type to what it is
right now. For example, right now, providing a MAC address in the
@00:0c:29@ range makes the ESXi server generate a new MAC address on
define.
I can imagine that some people outside always provide the same
MAC address to define their machine and rely on libvirt/ESXi to generate
a new one in that range for them.
This doesn't sound right. If they provide a MAC address we should
guarantee it, if they don't provide address we should generate one, that
suits hypervisor's needs.
Pino, you have more experience with ESX than I do, do you know whether
we could do the switch to mark MAC addresses static/generated according
to @mac_generated?
Michal