In a world where WF starts in 2 seconds on a 2 year old laptop, the default VERIFY_SUSPECT timeout of WF (currently 5 seconds) can easily create the condition where a killed and restarted server shows up in the topology before the old server leaves. Ideally, the VERIFY_SUSPECT timeout should not be longer than the time it takes to start the server process. A timeout of 1 second should suffice.