A simpler approach could consist in a flag stored in the cinder db to indicate if the volume has been formatted before.
Then we could handle the volume attachment procedure checking this flag's value first.
If the device has been used but Nova would not be able to mount it, we could return an exception inviting the cloud administrator to check for the volume integrity.
What do you think about that?
A simpler approach could consist in a flag stored in the cinder db to indicate if the volume has been formatted before.
Then we could handle the volume attachment procedure checking this flag's value first.
If the device has been used but Nova would not be able to mount it, we could return an exception inviting the cloud administrator to check for the volume integrity.
What do you think about that?