Is using JBoss Cache as distributed state repository a good idea?
Can JBoss Cache be applied in situation when you need the guarantee that each time you read something from repository you get the newest version of it? - and irrespective of which node in a cluster we consider?
You need to be careful with this stuff, but I would say yes. JBossCache is transactional, and can perform asynchronous or synchronous atomic operations on the distributed state. I've had good experiences with it, and can recommend it. It can be temperamental and obscure at times, but I've yet to find anything else that can do the same job.
With Hazelcast, it is always guaranteed. Reads return always the newest version, regardless of the cluster node.
Hazelcast is open source (Apache License), transactional, distributed caching solution for Java. It is a little more than a cache though as it provides distributed implementation of map, multimap, queue, topic, lock and executor service.