I've brought up hibernate using oscache and I'm having a hard time getting it to work within a cluster using the JavaGroups listener.

My cached data is never being invalidated between VMs and I think, from what I describe below, that they aren't registering with each other in the broadcast group.

Here's what I'm seeing (or not seeing):

If I start one VM, it prints a message for it's own broadcast listener being added to the group:
"14:01:56,601 INFO JavaGroupsBroadcastingListener:166 - A new member at address 'MarcusPopetz:3350' has joined the cluster"

When I bring up the second VM, it prints a similar message to the log file with a it's own slightly different port number but nothing about adding the new member shows up in the 1st vm's log file.

I then load a cached object in one VM (and it IS being cached because I've lowered the log level for both hibernate.cache and the opensymphony packages to check), change the object and call hibernate.update, hibernate.flush, hibernate.close

The other VM never sees the change and continues to use it's stale version of the cached object.

Does anyone know what sorta log4j settings I have to have to see messages about broadcasting stale object announcements or have any other ideas to figure out what I'm doing wrong?

If it's important, I'm running both these on the same machine. (I don't know much about broadcast protocols)

Thanks,
Marcus

We don't intend that you would use OSCache as a clustered cache. Use SwarmCache or TreeCache for clustering.

Given that OSCache supports clustering, what was the reasoning behind that? I'm curious because based on the research I've done so far OSCache seems to be more flexible and more mature than SwarmCache.

I'm also interested in this subject. I'm a fan of OSCache and it does support clustering through clustered flushing. Seems like OSCache would handle the flushing outside Hibernate, so is there a reason why it would not work?