I am doing a "flood test" of an MDB
using the new JDBC2 PersistenceManager
and getting some very, very odd results.

I am sending 10 messages into a Queue,
in less than a second, but my MDB only
fully processes maybe six of seven of
them--sometimes not even five are fully
processed.

I have placed try/catch wrappers around
every conceivable failure point in the
MDB, but the MDB just "dies" sometimes.

I sincerely believe that something is
dreadfully wrong with JBossMQ as I have
traced the hashcode's of each instance
of the MDB's that "dies" (but no runtime
exception is ever thrown by anything!)
comes back to process other messages
without being dumped as I thought was
what the specification states *should
not* happen, except in the case of a
failed transaction--not in the case of
a runtime exception occuring.

Has anyone had any similar problems with
sending many messages at once to their MDBs?