Migrations timeout causes a lot of migrations jobs to be spawned simultaneously
When the timeout of 300s is reached (which is very conservative) we throw an error. This error causes a retry which tries to create a new migration. We end up with thousands of migrations jobs most of them pending and some running in the same time.
Proposed solution: Add a property to the CRD to store the name of the last running migration job. The operator should not attempt to create a new migration as long as the previous one is running. Even if it attempts to create a new one it should make sure to terminate the old one properly before creating the new one.
cc: @twk3