I've tried to make this test more robust to timing issues. Basically I've multiplied the times between events by 10 so each "tick" of the test is now 5 seconds. I also modified the test such that it does its assertions "halfway" between the events. e.g. Time - event 0.0 - notification 1 2.5 - check notification 1 5.0 - notification 2 7.5 - check notification 2 etc. That should hopefully avoid the assertions happening at about the same time as the events themselves (or more accurately with the previous code the next one) and therefore randomly seeing the wrong state. -- xxxxxxxxxxxxxxxxxxxxxxxxxxxx Adrian Brock Chief Scientist JBoss, a division of Red Hat xxxxxxxxxxxxxxxxxxxxxxxxxxxx