Comment 3 for bug 1683038

I wonder why an openstack project would not configure images built to run on openstack to use the openstack metadata service. Is there a reason? Generally speaking I think it should perform much better as the openstack metadata service is often quite slow. Cloud-init crawls the metadata service and on openstack that will take probably 6 requests while EC2 will take several more. (Openstack provides lots of data in json format in a single payload while Ec2 provides individual bits in individual urls).

All that said, I'll consider it a bug that if you have only 1 or 2 entries
in the datasource list that Ec2 strict should be assumed.