On 28/06/16 04:06, Crag Wolfe wrote:
> I agree that will fix the template test. It doesn't fix the issue that
> template test raises though if a user happens to create a template
> similar to the current one. If we want that kind of template to fail
> validation, maybe that should be another bug. I'm not totally sure it
> should be considered an invalid case though -- the user may not really
> care which subnet the server gets associated with if there are multiple
> subnets, and the user may want to create the network/subnets/server all
> in one template.
>
I still think we should fix the bug, but I also think we should signal
to the user to avoid the network property by deprecating it.
And I think the tempest test should do what we recommend, which is
explicitly specify what subnet they want to connect to.
On 28/06/16 04:06, Crag Wolfe wrote: subnets/ server all
> I agree that will fix the template test. It doesn't fix the issue that
> template test raises though if a user happens to create a template
> similar to the current one. If we want that kind of template to fail
> validation, maybe that should be another bug. I'm not totally sure it
> should be considered an invalid case though -- the user may not really
> care which subnet the server gets associated with if there are multiple
> subnets, and the user may want to create the network/
> in one template.
>
I still think we should fix the bug, but I also think we should signal
to the user to avoid the network property by deprecating it.
And I think the tempest test should do what we recommend, which is
explicitly specify what subnet they want to connect to.