2 Replies Latest reply on Sep 9, 2013 12:57 PM by Wolf-Dieter Fink

    Clustering two server with different deployments

    slrslr Newbie



      I am using Wildfly 8.


      I have a question regarding clustering, for which I can find the good architecture type to use.


      Currently I use a domain (to benefit from the administration CLI, APIs and so on) with only one host.

      In this host, let's say that I have 5 servers : A, A2, B C and D.


      What I want to do is :

      - A and A2 can have the same applications but with different versions of the code.

      - The client targets only a single entry point (so a load balancer can be use like mod_cluster)

      - When A and A2 are up, the load is balanced between each other.

      - When one is down, the other takes the load

      The goal: having a backup server to be able to change the code version without having a service interruption.


      My thoughts:

      - Putting A and A2 in the same server group is not satisfactory because the boot/stop/deployment is done at "group level", i.e. the servers can not have their own life cycle and version of the code (except if we can deactivate farming, is it possible ?? )

      - I saw things about load balancing between several hosts. But if am not sure that having several hosts is related to clustering. In this case I would need to have several hosts on the same machine. Is it possible ? In this case, I do one host with A, B, C, D servers, and another host with only A2. That seems weird. Doing 5hosts with only one server on each seems weird as well.


      Do you have any other idea on how I can target with a single entry point two servers in a single host ?


      Thank you in advance.