1 2 Previous Next 16 Replies Latest reply on Sep 22, 2014 1:41 PM by jbertram Go to original post
      • 15. Re: Routing messages to runtime discovered hosts
        zulutime

        OK, I'll try to be more specific :

         

        Context :

        2 centrals servers (HA).

        Thousands of satellites servers messaging together and with centrals servers

        From time to time, some satellites are created and some are dropped (dynamic list of server)

        Producers and consumers everywhere.

        Satellites are on a not very reliable WAN.

         

        Question 1:

        Consumers host (satellite or central) may be down or unreachable, but the send must not fail from the point of view of the producer, message must be pesisted and sent asap. Is the best solution to accomplish that is to create a local queue, a remote queue and to bridge them?

         

        Question 2:

        Every Satellite is running the same webapp, so every one is consuming from the same queue definition. How can I specify at sending the host I want to send to ? Do I have to create a local queue/bridge for every host?

         

        Question 3:

        Central servers cannot have connection opened permanently with every single satellite (OS limitation). Can I configure bridge, or connection factory to close connections when sending queues are empties (with some keepalive time)?

        • 16. Re: Routing messages to runtime discovered hosts
          jbertram

          Question 1:

          Consumers host (satellite or central) may be down or unreachable, but the send must not fail from the point of view of the producer, message must be pesisted and sent asap. Is the best solution to accomplish that is to create a local queue, a remote queue and to bridge them?

          If the send "must not fail from the point of view of the producer" then I don't see you have any choice but to create a local queue, a remote queue, and to bridge them given your current architecture.  There may be other architectures which are better so I can't say whether or not this solution is the best solution.  If you have to maintain a queue and a bridge on every host used for communicating with every other host that is likely to be a configuration nightmare.

           

          Question 2:

          Every Satellite is running the same webapp, so every one is consuming from the same queue definition. How can I specify at sending the host I want to send to ? Do I have to create a local queue/bridge for every host?

          Yes, I think you would need a queue/bridge for every host to which you wanted to send messages.

           

          Question 3:

          Central servers cannot have connection opened permanently with every single satellite (OS limitation). Can I configure bridge, or connection factory to close connections when sending queues are empties (with some keepalive time)?

          We don't support that functionality exactly, but you can manually call start/stop on the bridge.

          1 2 Previous Next