In case something went wrong, how will you handle and control rollbacks?

I will recommit the previous version of the code to be the new and current version in my VCS. With this a terraform run will be triggered which will be responsible for running the old code. Don't forget that terraform is more declarative. Make sure that all the things provisioned in the code for rollback are contained in the old code. Ensure that it does not get destroyed on running the old code because of the lack of these. In case the state file got corrupted from a terraform run recently, then I will take the help of the State Rollback Feature of Terraform Enterprise for roll back to the previous latest state which was in a good state. This could be done due to the reason that every state change is versioned.