A cluster failover in the early morning Pacific caused some users to experience errors for several minutes (rather than the typical error window of several seconds or less) due to configuration changes made to mitigate recent Cloudflare routing issues.