Conincidentally, I suggested the same thing, check Brian's post (MCMP
SUSPECT) in the archives (this week).
Marek Goldmann wrote:
Proposing an enhancement for mod_cluster manager: I think that
mod_cluster manager should remove not running (terminated, crashed)
nodes from node list.
I'm using mod_cluster on EC2: I'm starting/shutting down instances.
After a few such operations I don't know which nodes are really
running and which are only ghosts.
Or maybe I just misconfigured mod_cluster?
Lead JGroups / Clustering Team