On February 1st, starting at 7:45 AM, users belonging to partner tenants encountered authentication issues. The problem surfaced only when users from partner tenants attempted to access those tenants other than their own, situated on different clusters. The engineering team identified the issue at 9:06 AM CET, attributing it to recent changes. Swiftly responding, a fix was implemented and deployed across all clusters by 10:40 AM CET, successfully restoring normal system functionality for the affected users. To prevent future occurrences, we are enhancing our automated testing procedures, ensuring the robustness of cross-tenant functionality and averting similar issues.