7 Replies Latest reply on Mar 17, 2016 2:10 AM by Udit Mishra

    How to ensure failover while connecting to the endpoint of a cluster?

    Udit Mishra Novice

      Suppose my cluster consists of 11 nodes, out of them one is DC and the rest 10 are servers(also HC), each one is on different physical machine.

       

      While connecting to the cluster(from hotrod client) its sufficient to give any node's address out of the 10, Now suppose I gave address of node 4, and before reaching the request, node went down. What would be the outcome on client side?

       

      Is there anything through which we can ensure that the if the initial request for connecting to the endpoint fails, it can re request to some other node???  Like if all my nodes are addressable and I have a list of them, and if I am not able to connect to node1, I would say, pick the next node from the list and try with that. It should be just for the initial connection. Since, once I am connected, then I know the topology, I am topology aware.

       

      Alternatively can I put Apache in front of the servers?? so that my hotrod client connect to the server and if its down then Apache should redirect it to some other node