Core Bridge Fail over to nodes on a different cluster

[zeebot]

Can someone please help me -

I have two seperate clusters each with two nodes.

Cluster 1 nodes sends JMS messages to Cluster 2 nodes, however i woud like to ensure that each node in Cluster 1 can failover to the any of the nodes in Cluster 2.  Basically i want mulitple failover - failback between two seperate clusters.

Scenario

NodeA on Cluster1 sends JMS message to NodeA on Cluster2

NodeB on Cluster1 sends JMS message to NodeB on Cluster2

NodeA on Cluster1 goes down / shutdown - NodeA on Cluster 1 fails over to NodeB on Cluster 2.

NodeA on Cluster1 is started - NodeA on Cluster 1 fails back to NodeA on Cluster 2.

I am using Jboss 6.1.0 and HornetQ 2.2.5.Final - please can someone help with what the best configuration is.  I have tried using a Core Bridge but cannot seem to get it to failover.  Searching has found this issue HORNETQ-769 - do upgrade to 2.2.8 or use the workaround.

Is Core Bridge the best solution ? or is there another configuration that can help me.

Kind regards

Zeebot

[ataylor]

your scenario doesnt really make sense, first of all you will need backup nodes for each live server, its not possible for a live server to fail over to a nother live server.

[zeebot]

Hi Andy

Ok so would having a backup server for each live server help resolve my issue for ensuring the JMS message was sent successfully to the second cluster ?.

Also lets forget the notion of clusters as i am not so worried about clustering as i am about ensuring the JMS message is either sent to the correct server in the event it had crashed / shutdown.  So NodeA wants to send a JMS message to NodeB on another network, however if NodeB is not in operation NodeA will send to NodeC or possibly even Node D if NodeC is not available also.

I could write an application to deliver the message to the other consumers but want to know if JBOSS does this out of the box without having to spend too much effort.

Any help would be much appreciated as i am struggling to setup the configuration with JBOSS 6.

Kind regards

Zahid

[ataylor]

Ok so would having a backup server for each live server help resolve my issue for ensuring the JMS message was sent successfully to the second cluster ?.

Yes, but remember backup servers are passive and dont become live until failure occurs. You can configure as many backups as you want but they will all be passive, iwould start by looking at some of the failover examples shipped with HornetQ

your topology would look something like

node A(cluster1) ----------------------------> live A(cluster2)

                                            |

                                            |-----------backup A(cluster2)

node B(cluster1) ----------------------------> live B(cluster2)

                                            |

                                            |-----------backup B(cluster2)