I do not know how to support my product. Suppose that I have a bad network. The jChannel instances bind to a bad address where data cannot be routed to other cluster members. I am using JDBC_PING. All jGroups nodes successfully attach to the database, so JDBC_PING works perfectly. But, the 4 members of this cluster form 4 separate clusters.
No error is logged anywhere. If I enable full TRACE debugging (via at the bottom of the stack), I see nothing to indicate there is a problem.
I am not stating that I want jGroups to work on my bad network.
I must be able to determine that the computers have joined a cluster. I must be able to report a failure if they haven't. Is there a solution to this problem? I have looked into this a great deal. I am thinking that my only solution is to have all nodes communicate to one another via the database and using my own solution, but this is silly since jGroups already has all of this information!