Behavior of Ensemble following a system failure and restart
This paper describes how Ensemble behaves after a system crash or similar uncontrolled shutdown or failover. Ensemble is normally configured to start processing messages automatically when the operating system restarts or when the system has failed over to a cluster member or mirror member. No manual intervention is required. But, in order for your system to robustly handle system failure and restart, you must understand how Ensemble handles these conditions and develop your productions using the rules and guidelines in this document.
The behavior in the event of a restart is very dependent on the solution and the participating applications. Ultimately, it is the responsibility of the solution designer to address each possible failure mode.