Despite the already existing self-defense mechanism that monitors the Active MQ queue sizes, it may still happen that a few message cause OOM close situation in case the messages are huge.

In order to keep Wilma stability even in such exceptional cases, we need to monitor the percentage of the used memory, and act in case AMQ use an unsafe level of memory, and restart the AMQ, to reset its memory usage. What we loose: some (huge) messages won't be logged - what we win: Wilma will survive killer size messages too.

Of course this mechanism will be activated only when the allocated/available memory size is not enough to handle the amount and size of the messages (so other solution is to increase the allocated memory).