11:15pm Pacific — Our engineers noted earlier this evening that one of the database servers used for greylisting was not receiving updated data correctly.   This  would have impacted only those customers that were both serviced by that particular database cluster, and that had greylisting enabled, and would have resulted in a small quantity of messages being deferred multiple times instead of once.   The issue has been resolved, and any messages that were deferred should have been subsequently accepted and processed normally.