1 of 1 people found this helpful
I don't know what version of JBossAS you use, but if you use 6, you could use distribution (Infinispan: DIST) and requests landing on an incorrect node would get forwarded to the right node, which has the data.
You could also use synchronous replication (Infinispan: REPL_SYNC), which blocks the caller until the modification has been shipped across the entire cluster. Then, if the next request hits a different node, it will still get the correct data (e.g. a read).
I am using jBoss 4.2.3 and jBoss 5.1.0. What about this http://docs.jboss.org/jbossas/docs/Clustering_Guide/beta422/html/clustering-http-app.html
Is it going to help? Also, is there any configuration guide to achieve this.
Let me know if you need more information.