A couple of quick notes:
* Finding instances for new nodes means that Databricks is attempting to provision the AWS instances necessary. This will often take longer if A) the cluster is larger, or B) the cluster is a spot cluster.
* They should not stay in this state for longer than ~30 minutes however (because of a timeout); if it does please let us know.
* It is on the roadmap to A) allow users to terminate pending clusters, or B) have Databricks return a smaller cluster if it wasn't able to provision all the instances, but that's not there yet.