Honeycomb performed scheduled maintenance on May 4th (EU) and May 7th (US) to replace their decade-old Kafka cluster, during which data ingestion continued but remained unqueryable for roughly 25 minutes in EU and 1.5 hours in US, with no data loss. The report identifies inadequate advance customer notice, imprecise language about impact scope, and the absence of a disruptive-maintenance coordination process, and outlines new protocols for longer internal coordination timelines, clearer accountability, and wider customer communications.