The issue has been resolved. The root cause was an automated recovery following a disruption to one of our microservices.