A couple of questions...
- What version of JBoss 7 and HornetQ are you using?
- Can you attach your configuration from both cluster nodes?
The version of JBoss i'm using is EAP 6.2. The used version of HornetQ is the one embedded in JBoss, i think it's 2.3.12 as far i can tell from looking at the jars in the JBoss modules directories.
I have attached the hornetq part of the configuration from both cluster nodes as file. One of the difficulties regarding this problem is, that it occurs so rarely. Right now i have two JBoss nodes running (on two different machines), on which i can reproduce the described error reliably (i keep them running and tell everyone not to restart the system so i can do sth. about examining it). So i would very helpful for any hints how to examine the state of the clustered Hornetq servers or the topic via CLI, REST, JMX or whatever.. I couldn't figure out, for example, what could be a way to find out whether the hornetq servers are correctly linked at the moment or not.