The volume needs to be "in-use". I see you are resetting it's state to "available". When in-use, cinder will attempt to notify nova to refresh the size of the disk. See:
On iSCSI backends the volume needs to first be refreshed on the host before the VM can be made aware of the new size. I suspect there may be some operation that needs to be done on RBD backends as well, but I cannot say for sure.
To reproduce, I recommend you use any iSCSI backend. We used a Dell Unity, but any iSCSI backend will do.
Hi Samuel,
The volume needs to be "in-use". I see you are resetting it's state to "available". When in-use, cinder will attempt to notify nova to refresh the size of the disk. See:
https:/ /github. com/openstack/ cinder/ blob/master/ cinder/ volume/ manager. py#L2939- L2943
On iSCSI backends the volume needs to first be refreshed on the host before the VM can be made aware of the new size. I suspect there may be some operation that needs to be done on RBD backends as well, but I cannot say for sure.
To reproduce, I recommend you use any iSCSI backend. We used a Dell Unity, but any iSCSI backend will do.