Implement a first simpler reliability strategy only leveraging persistent object store. When a session is recovered using this strategy the objects present in the stores are flushed into their respective entry-point (maybe with an update instead of an insert since the object are already in the store, or we can recreate the stores from scratch and perform a normal insert, to be investigated) issuing  a single fireAllRules at the end of this flushing process with an agenda filter that prevents the execution of all consequences.