Microsoft explains unusual server problem prolonging global Outlook and Teams outage

A major outage struck Microsoft's suite of services on Thursday, crippling Outlook, Teams, and Copilot for users worldwide. The disruption persisted for over five hours in some cases, with reports piling up on Downdetector numbering around 138,000 complaints in the US at its peak. Although Microsoft initially identified issues in the Americas, reports soon spread globally, with Service NSW in Australia warning of unavailable transactions.

Major services like Exchange Online, Teams, SharePoint Online, Copilot, Defender XDR, and several Purview services were reportedly affected.

Microsoft attributed the incident to a large number of servers unexpectedly becoming unavailable, overwhelming the remaining infrastructure and taxing the software responsible for establishing secure connections. Additionally, recent configuration updates were being investigated as potential contributors, and all deployments and changes were paused during the review.

The recovery process faced challenges, as restored servers sometimes became unresponsive once again once they received traffic. Microsoft attempted to restart unresponsive servers, reroute traffic, and temporarily limit affected traffic in certain locations. They also tried rebuilding some servers on a previous configuration to test if that would alleviate the issues.

The outage proved peculiar, as Teams' core chat and online meeting functions remained largely functional, but scheduling meetings or accessing the calendar often failed. This was attributed to Teams relying on Exchange Online, the service responsible for storing and delivering work email and managing calendars. Similarly, ChatGPT and Claude from OpenAI were affected as the Microsoft 365 connectors that allow AI assistants to access users' email, calendar, and files with permission were impacted.

OpenAI and Anthropic both reported failures in new connections and certain functions, causing problems with email, calendar, and search requests in Claude inside Outlook. Microsoft's Copilot also struggled to access work email, calendars, and files during the outage.