I have a cluster of three JBoss 4.0.2 servers running on linux with JBossMQHA running.
I have three queues deployed as service (SARs) on each node in the cluster; with MDB consumers for the queues.
This setup has been running succesfully with a steady throughput on each queue for around 3 months.
Under normal circumstances the queues on the master node are in a state of 'Started' (3) (When viewed via JMX console) and the queues on the non-master nodes are in state 'Registered' (8)
Today each queue on each node stopped being available and when checked had become in the state 'Destroyed'. I can find no documentation or reference to this problem and there is no exceptional/error output in the log files.
The only way I have managed to return the queues to the correct state is to re-start each jboss instance.
Does anyon know:
1) How this situation would come about?
2) How to prevent it in future?
3) How to place a queue back into 'Started' from 'Destroyed' without restarting jboss?
Any thoughts on this guys? The problem has re-occued today...