I’m not sure if you can switch from single to double, and then back to single. I don’t remember if this configuration was available when I was testing fdb, or if you just increased the number of processes in order to scale the cluster.
I do remember getting into a state where the status said it was migrating data, but there was no available node to migrate it to (because I wanted to shrink the cluster). Effectively, the cluster was deadlocked.