The system is now fairly stable. I’m continuing to run processes to locate and merge data objects based on certain identifier values. These are run in batches of 50,000 identifiers every 30 minutes or so, and result in momentary slow search performance.