I have a two-node cluster setup where one node holds the Master and the other a Read/Write copy of the replica. NDS is not segmented and has only one replica. The problem I'm facing is that if the node with the Master replica goes down, the services migrate over but, no one can log in until the Master replica node is back up? My understanding was that you only need a Read replica to login?

Also, I've noticed when the node with Master copy is high in utilization, other services are effected on other servers? For example we have an accounting software, MAS90, on a third server. If the node with Master replica is busy, then MAS90 may have communication errors? This might not be related but, that seems to be the trend?

Thank you,