Databricks being a managed services offering, its designed to abstract the complexity of having to manage the finer networking details of your underlying infrastructure such as IP addresses. You can specify a CIDR range for the subnet within which databricks compute is spun up and databricks manages the assignment of IPs for simplicity
I would not recommend doing this, I would however try to find a way through the init script to register the existing cluster IP in private hosted zones DNS (if AWS), so your apps will not be aware of the IP, but always refer to a hostname.
