So, the reason why this becomes relevant is: If Glance deployments are expected to allow setting multiple location on such drivers, we need to communicate to the operators on the right strategy to be used when doing so. Mentioning the severity of risk is important as well.
Hi Andrew,
Thanks for the input. Currently, Nova is relying on custom location logic for Ceph specific snapshots https:/ /github. com/openstack/ nova/blob/ 824c3706a3ea691 781f4fcc4453881 517a9e1c55/ nova/virt/ libvirt/ driver. py#L1517 .
As a part of the v2 transition, multiple locations need to be enabled https:/ /review. openstack. org/#/c/ 301741/ 8/specs/ newton/ approved/ use-glance- v2-api. rst@260 .
So, the reason why this becomes relevant is: If Glance deployments are expected to allow setting multiple location on such drivers, we need to communicate to the operators on the right strategy to be used when doing so. Mentioning the severity of risk is important as well.