A configuration change caused an unexpected shift of traffic between clusters, which manifested as API failures to end users. The impact window was from 16:50 -> 16:58 PT, with traffic recovery starting at 16:55 PT. We do not believe this issue will recur, and will be taking proactive steps to ensure such an issue does not recur.
Posted Jan 19, 2023 - 17:30 PST
Monitoring
A fix has been implemented and we are monitoring the results.