TCPCacheServer as a single point of failure?

[superflav]

I'd like to set up a pair of TCPCacheServers that are clustered together and then refer to them in each of my JBoss AS instances as primary and 2ndary TcpDelegatingCacheLoaders. I have no problem chaining them together in the CacheLoaderConfiguration, but then when I take either one of the TCPCacheServers down, my JBoss AS instances go haywire. Is this even possible? If not, are there any recommendations on how we can prevent a TCPCacheServer instance from being a single point of failure?

Thanks.

[manik]

This is a feature on our roadmap. Sadly for now, taking a TCPCacheServer down means first taking the cache down. :/

JBCACHE-1260