Resolved - Between 12:35-12:54 UTC (04:35-04:54 PST) we saw elevated numbers of 5xx responses to API requests, and elevated response times to API requests. The problem was caused by unrecoverable errors on an upstream provider's API which caused API worker processes to recycle at high rates, in turn leading to request queues building up and sometimes time out.
Jan 15, 04:30 PST