i'm using Weblogic6.1 sp 2. Is it possible to do clustering and load balancing for Message Driven Bean (MDB)? I have deploy a MDB to a cluster containing 2 managed server. I try to make this clustr as the consumer for a destination queue so that these two managed servers can do the load balancing by itself. Can someone tell me how to configure the these two managed server to consume messages from a same queue? Is it possible to do so? Am i having the right concept? Thanks you
MDB clustering is really about JMS clustering. Unfortunately WebLogic 6.1 does not support JMS Clustering. To be perfectly honest even 7.0 is shaky in this area. JMS is a hard service to cluster and this is the weak point for most application servers. In 6.1 you can create the JMS Queue on one node and target it to the cluster. This is still a single point of failure for that queue but it is the best 6.1 offers. See this link: Configuring WebLogic JMS Clustering in WebLogic 6.1 7.0 is a bit better with its use of Distributed Destinations and Migratable Services but it is still lacking. See this link if you are interested: Configuring WebLogic JMS Clustering in WebLogic 7.0
Hi, I feel that when you talk about EJB clustering and when you talk of JMS clustering you refer to two different things. Thus when we talk of MDB clustering we do not talk of JMS clustering. I think that MDB's can be clustered as normal beans. Their onMessage is triggered when the Queue or Topic receives a message. Since we use a Factory object to look up the Queue or Topic configured at the deployment descriptor through the following configuration should be possible 1) Server A has a Queue 2) Server B and C has the MDB's Now what happens when server A has gone down is that the JNDI lookup for the Queue or Topic is no longer available and hence the MDB's will raize exceptions. But lets say that 1) Server A has a Queue (JMS Server with Queue1) 2) Server B and C has the MDB's. 3) Server B has another JMS Server with Queue1 deployed on it How the weblogic server is optimised to handle such kind of configurations is unknown to me.
I still say that it is pointless to talk about MDB clustering without talking about JMS clustering. You can cluster the hell out of your MDB components but if JMS is still a single point of failure, you lose.
You showed up just in time for the waffles! And this tiny ad:
Building a Better World in your Backyard by Paul Wheaton and Shawn Klassen-Koop