Hi,
We have problems with NLB settings in RDS 2012.
The situation is as follow:
We have 3 terminal servers 2012 (2 session host) and 1 (connection broker)
We have 1 collection and in it are the two session hosts
Each session host has two network cards. two subnets (172.30.Xx and 192.168.x.x) and the session broker with card 1 (192.168.x.x) (Now it's just 1 after doing many tests but before it also had 2 as the host session )
in test situation, when using only one subnet, the collection and the session broker and RoundRobin (without NLB) we have no problems. But it is necessary for us to have 2 subnets and one NLB cluster.
In the subnet 192.168.X.X we configured an NLB cluster ip 192.x.20. The cluster is multicast IGMP. The cluster with some configurations has been unstable but now is not.
Our customers are Thin client. When we are going to connect to the cluster IP (192.168 X.20)
What we see in the logs are all these situations:
1. Sometimes connects to the cluster IP (192.168.x.20). It connects quickly.
2. Sometimes connects to the "real ips" of the host but the 192.168.x.x subnet, NO to the cluster IP. It connects overall good.
3. But sometimes appears in the logs that it connects to another subnet IPS that have nothing to do with the cluster, that is the actual host IP subnet but (172.30.x.x). In this case the system has many delays when connecting.
Even 1 min
Although the thinclient I configure to be connected to 192.168.x.20 (IP of the cluster), it really connects randomly where it wants and how it wants ...
Thank you very much for everything.