Rancher 2.0.8 UI showing 503 error

I have a 3 nodes HA type 4 load balancer cluster setup with all 3 nodes acting as etcd, control plane and worker. As a HA test when am bringing down Node1 rancher UI showing 1 node1 unhealthy as expected…same thing for Node2. But when am getting only Node3 down am getting 503 timeout error on rancher UI.
Could someone please help on why this is happening? Am using Rancher 2.0.8