We experienced elevated error rates for synchronous API calls for around 30 minutes. We were alerted, intervened and are now still pursuing a root cause.