We are using max_over_subscription_ratio based on free capacity because we could have other data being stored on the same pool, and then over provisioning would be pure chaos.
I agree that we should consolidate the way we do calculations, but first we would need to make a statement on the assumptions that are made regarding the use of the pools used for cinder (ie: exclusive use by Cinder, and by 1 and only 1 service).
We are using max_over_ subscription_ ratio based on free capacity because we could have other data being stored on the same pool, and then over provisioning would be pure chaos.
I agree that we should consolidate the way we do calculations, but first we would need to make a statement on the assumptions that are made regarding the use of the pools used for cinder (ie: exclusive use by Cinder, and by 1 and only 1 service).