We experienced an increased error rate from the API and Dashboard. The cause was reaching a connection limit on one of our Redis databases while rolling out a new backend service.