Which method can be used to correct database replication issues in a cluster running Cisco Unified CallManager 5.0?