6:45am Pacific / 1:45pm GMT — Our operations team has removed the problematic database cluster from production, and message processing is returning to normal.  We will continue to post updates here.