Users began experiencing intermittent request failures starting at approximately 6:32am PDT. Omni engineers replaced a single instance of a service that had become unresponsive, and operations returned to normal by 7:05am PDT. Work is ongoing to prevent this issue from recurring.