Can anyone point to information on Message Bean clustering strategies with particular reference to Weblogic?
Clustering of MDB's has greatly been simplified in WLS 6.1.
There was a one to one relation between the MDB and the JMS server.
The MDB and JMS server deployment on the WLS 6.0 had some limitations which would be too much to discuss here.If you need some help on this I can reply to that in detail in separate thread.
As of WLS 6.1 this is what is to be done.
1) IN WLS 6.1 , JMS Server now can be deployed only on one WLS server instance within a cluster.
In WLS 6.O the JSM server could be deployed on a cluster i.e atleast it could be targeted the cluster.What happened internally was that the JMS server would be bound to the local JNDI tree and only one instance was availbale on the cluster wide JNDI tree.
2) In WLS 6.1 the MDB's now can be deployed on the cluster.i.e these MDB's can even get deployed on a WLS instance within a cluster on which the JMS server has not been deployed.If the JMS server is not available on the cluster at the time when the MDB gets deployed,the server log will issue a warning stating that the JMS server hosting the destination for the MDB is not currently available and that the MDB will connect to the destaination as soon as the JMS server get available anywhere on the cluster.
IN WLS 6.0 this was not the case .The MDB cluld be deployed only on those WLS instances on which the JMS server was available.
This is the summary
WLS 6.0 WLS 6.1
JMS connection | Cluster | Cluster
JMS servers | Cluster | one WLS server instance
MDB's | Cluster | Cluster
WLS 6.x JMS clustering is a misnomer.All that it gives is load balancing through the JMS connection factories.i.e the JMS producer from any WLS instance could reach the JMS server hosted on one WLS instance within the cluster using
the JMS connection factories.
There is no failover support in WLS 6.1 (but it could be acheived in WLS 6.0 using a slight tweak)
I hope this helps