Hmmmm, what's the rest of the stacktrace?
To be honest, you might be better off having a client/server arquitecture rather than requiring all your clients to talk UDP to each other... https://docs.jboss.org/author/x/2AU5
I do not have the full stack trace on the Java Console... all i see is repeating:
Mar 20, 2012 12:37:20 PM org.jgroups.logging.JDKLogImpl error
SEVERE: failed sending message to cluster (68 bytes): java.lang.Exception: dest=/22.214.171.124:5555 (71 bytes), cause: java.io.IOException: Invalid argument
I'll try to setup JGroups logging to show more, would you have any pointers on that?
What i am trying to achieve is a "Fault Tolerant SaaS", please let me explain:
Suppose a application for a restaurant, offered as SaaS (no installation, server, backup, etc...).
The problem is that if the internet goes down, the restaurant stops.
To solve that, i'd like that all clients in the restaurant (waiter, kitchen, cashier, etc) had all recent events and keep operating.
If i could run a infinispan cache inside an applet or java web start, that seems to do the trick....
Promoting one of the clients to "server" does not seem to cut it, as they would be subject to the same limitations.
Take https://github.com/infinispan/infinispan/blob/master/core/src/test/resources/log4j.xml, add TRACE for org.jgroups package and pass the location (path) of the log4j descriptor via -Dlog4j.configuration=file:/...
It actually works when setting preferIPV4Stack=true
Unfortunately, setting that seems to be a bit quirky in JNLP, so i opened a question about it: