We have a two server Windows 2019 cluster for Remote Desktop Services. These two machines are hardware based and they rely on a virtual Windows 2019 server with MSSQL 2019 to control the cluster.
Overall these two machines service the cluster fine with one exception. What happens with cluster services doesn't happen within the RDS system.
For example: if the cluster determines that it needs to balance the servers by moving a VM to one server or the other the RDS system doesn't seem to be aware of this. When you look at Hyper-V Manager it sees the VMs properly. When you look at the Cluster Manager, it too sees the VMs properly. However when you look at Server Manager and dig into the RDS Collections and look at where it believes the VMs are it sees the VM on the original server and in an "unknown" state. At that point if the user is still on the VM they will continue to work. If they disconnect and attempt to reconnect they won't be able to since the RDS side of things doesn't know where the VM lives.
How to we resolve this problem? This usually shows up on reboots however I have had to lower the balancer from high to medium since this problem has been seen to occur at times when balancing is needed.