Microsoft Weekly — November 09–13, 2026

A server under pressure fails its health checks. Automated traffic management removes it from service, pushing even more load onto the remaining infrastructure. What was designed to protect the system becomes part of the failure itself. In this episode of Support Engineering Weekly, Alex and Maya examine the Microsoft 365 incidents reported between November 9 and 13, 2026, starting with a major multi-service disruption affecting Exchange Online, Teams, SharePoint, Copilot, Defender, and Purview. We unpack the engineering mechanisms behind the cascading failure, including overloaded TLS handshakes, automated health checks, the thundering herd effect, and the decision to extend Exchange Online message timeouts to 48 hours to protect delayed email from unnecessary delivery failures. Beyond the infrastructure outage, we investigate how distributed-system behaviour creates unexpected problems across Microsoft 365. We explore why Copilot Chat can display outdated conversation history even when the underlying data remains safe, how a latent state-management defect can surface during an infrastructure migration, and why a seemingly minor SMTP header change can trigger duplicate messages and downstream support tickets. We also examine an Exchange Online quarantine failure that raises an important security question: what happens when a system designed to isolate suspicious messages cannot determine where to store them? The discussion then moves to Copilot's integration with Edge for Business, including the interaction between browser-based writing assistance and data loss prevention policies. Finally, we look at the connection between front-end behaviour and back-end functionality through two unusual incidents: restored Microsoft Teams appearing invisible to users because of a client-side rendering issue, and Word for the web failing to open its audio-file selection dialog under specific Chromium browser window conditions. Across these incidents, one lesson keeps returning: the visible symptom is not necessarily where the failure begins. For support engineers, administrators, and IT professionals, understanding the dependencies between infrastructure, application state, security controls, and user interfaces is essential to troubleshooting modern cloud services. Chapters 00:00 Introduction: When Recovery Makes It Worse 01:33 The Cascading Infrastructure Outage 06:11 Copilot Caching and Hidden State Bugs 10:14 Copilot in Edge and DLP Boundaries 11:35 Exchange Routing and Quarantine Failures 16:18 Teams and Word: Front-End Failures 20:51 Lessons for Support Engineers Learn more: Visit the Support Engineering Blog at https://blog.sadhan.ch/ for the full articles, technical references, and additional support-engineering resources behind the discussion.