Hello, I have look a bit more into the defect. The stacktrace is showing some interesting bits of information : http://logback.qos.ch/jenkins/job/logback/ch.qos.logback$logback-core/298/te... JUnit is timing out right when the main thread is in the fail method of the last AssertTrue (L105 of AsyncAppenderBaseTest) I successfully reproduce the stacktrace on my computer at home (in 50% of test runs) for a specific test code outside of logback. But it is not reproductible on my laptop at work. So I still have to work on it to discover the full story. We can probably already assert the following : - Since we did enter the fail function. The assert did fail. Why? I think this is the main question - Is the 2sec timeout of the test was reach? How? There is 1sec dedicated for the join but the test length is only 1ms. By the way, it is not related to the bug, but I don't see any constraints that ensure that the worker thread is block by the BlockingQueue when the main thread try to interrupt it. The worker thread might be in the RUNNING state. Am I wrong ? Maybe it is expected ? Thanks a lot for your feedback ;-) Best regards, Cyril