Maintenance of the database clusters in the central storage facility to optimize performance caused a temporary caching issue where replaced IPs were still in the pool. Cache purge has solved the issue. Possible 10 minute downtime in affected regions.