One other observation....after seeing the lock contention warnings for a few minutes, that instance receives an EXIT message and leaves the cluster. All of the other boxes then start displaying warnings about discarded NAKACK messages from the locked instance, and then finally the locked instance rejoins the cluster and the locking issue seems to be cleared up. Don't know if this leaving and rejoing the cluster is a symptom or related to the cause....
I have the same problem in my server, but, I don't use cluster.
sometimes, I delete the records from the table jms_messages, and after some time, the problem continues
how do you solve this problem??