From about 5:30 am until about 14:20 PST some customers experienced intermittent timeouts. The reason was twofold: 1) a customer with demanding work-loads should have been assigned to dedicated hardware resources but was colocated with the freemium users on infrastructure scaled for freemium plans. 2) this customer had a runaway batch job distributed across data centers forming an unintentional DoS attack. The immediate problem has been eliminated, and we have taken steps to make sure customers with big plans are always assigned appropriate infrastructure.
Posted Feb 12, 2019 - 23:37 CET
This incident affected: api.sanity.io and apicdn.sanity.io.