api call should fail on request it cannot fulfill

When the user tries to create an instance of an image where the root disk is larger than the size of the vm-type requested, the API method should fail. For example, I had a small which had a disk size of 1GB, but the image I wanted to launch was 1001MB when unencrypted. The instance would get into a running state (according to the cloud controller) but would terminate straight away.

The error message I got in the nc.log is:

error: insufficient disk capacity remaining (505MB) in VM Type of instance i-39650626 for component root
the following and a running instance.the following and a running instance.

Although with hindsight, I can plainly see what this error is telling me, it really threw me for a while as I thought there was a problem with the kernel or ramdisk as the instance went to a running state.

When I use a medium, which is set to 2GB, I get a running instance successfully.

From the ImageCache table, I can see the image size in there, and I can see the maximum disk size in the portal, so I think that any request where the order is impossible to be fulfilled should be rejected at the API.

Robbie