We are using JBM 1.4.0.SP3 on JBoss AS 4.2.2.GA without clustering.
We have observed that (under high load) one suddenly dying remote JMS topic subscriber (e.g. when blue screening or disconnected from the network) might block message delivery to all other clients for a rather long time. It takes somewhere between ten to twenty minutes before finally an "java.io.IOException: Connection timed out" is thrown and processing continues normally. This can be accelerated by rebooting or reconnecting the failing client, respectively.
Is this the behaviour to be expected? Is there a timeout parameter that can be changed to reduce the blocking time significantly?
I have stack traces from the blocking server-side threads (one per JMS session), but as they are rather long I just wanted to get a quick opinion before posting them.
Thanks in advance for any help.