[Ovirt-devel] [PATCH server] First attempt at refactoring the vm_controller using the service layer.

[Ian Main]

On Mon, 20 Apr 2009 15:13:13 +0000
Scott Seago <sseago at redhat.com> wrote:

> This code is based on David Lutterkort's proposal outlined in
> https://www.redhat.com/archives/ovirt-devel/2009-April/msg00086.html

[SNIP]

> +  def svc_create(vm_hash, start_now)
> +    # from before_filter
> +    vm_hash[:state] = Vm::STATE_PENDING
> +    @vm = Vm.new(params[:vm])
> +    @perm_obj = @vm.vm_resource_pool
> +    @current_pool_id=@perm_obj.id
> +    authorized!(Privilege::MODIFY)
> +
> +    alert = "VM was successfully created."
> +    Vm.transaction do
> +      @vm.save!
> +      vm_provision
> +      @task = VmTask.new({ :user        => @user,
> +                           :task_target => @vm,
> +                           :action      => VmTask::ACTION_CREATE_VM,
> +                           :state       => Task::STATE_QUEUED})
> +      @task.save!
> +      if start_now
> +        if @vm.get_action_list.include?(VmTask::ACTION_START_VM)
> +          @task = VmTask.new({ :user        => @user,
> +                               :task_target => @vm,
> +                               :action      => VmTask::ACTION_START_VM,
> +                               :state       => Task::STATE_QUEUED})
> +          @task.save!

I can see this is a bit odd.. if we're calling this from QMF then
there's going to be issues here..  In the current API I don't have
create VM returning a task, which is kind of beside the point, but if
it did then we would have to have a QMF object of this task figured
out when the command returned (there are other APIs that do return
tasks like start_vm).  I'm not entirely sure how that would work using
this API right now.. we'd have to somehow figure out which task it was
that was put in the DB.. I wonder if there is a better way to approach
this?

I'm not going to argue this hard here because I know this is a first cut
and just a refactoring but I do think there should be no tasks
associated with creating a VM definition for the API.  I think we've
talked about this before.. not a big deal, just wanted to bring it up
for future reference :).

	Ian