Our goal is to provide stable solution that will protect us from most common failures eg. one JBM node failure or network failure.
We have clustered JBoss Messaging 1.4.8 and standalone application server JBoss 5.1.
On JBoss messaging we have clustered queue 'QueueA' deployed.
On JBoss 5.1 we have a MDB consuming from 'QueueA'.
We tested our configuration against node failure and it works fine.
After crash of one node via hard kill or machine restart MDB is connecting to the second node and consumes messages.
However we are puzzled by the following case: if we block the connection to one node, on the client machine, via iptables DROP we notice that MDB is hanging.
We tried with with very different bisocket and connection factories configurations but the problem remains unresolved.
Interesting thing is that the control connections to the node that is not blocked are closed immiedately after iptables block and client still tries to connect to blocked node.
Is it a known issue? Can we find somewhere proper messaging and client configuration?
Thanks in advance
Retrieving data ...