
Microsoft has attributed last week’s Microsoft 365 outage to a software defect that sat dormant for more than three months, then collided with a traffic surge and a capacity shortfall to take down over 80% of the servers in a key network gateway.
The disruption began on Thursday, Nov. 12, at 4:20 p.m. UTC and ran until 4:10 p.m. UTC on Friday, Nov. 13, in the timeline of Microsoft’s preliminary post-incident report dated Nov. 15. Exchange Online mailbox access through Outlook and other mail clients was the main casualty, along with delayed email delivery. The incident also touched Copilot, SharePoint, OneDrive, Defender, Purview, Fabric, Intune, Planner, Power Apps, To Do, Power BI and Universal Print. Teams chats and meetings largely kept working.
Microsoft described three factors. The first was a latent defect, a case of lock contention inside an internal Windows process that exhausted CPU and memory on gateway servers. The code came from a fix prepared only for Microsoft 365 networking gateways, and it does not apply to broader Windows deployments. The change began rolling out in July and spread globally over several weeks, running for more than three months without visible impact.
The second factor was load. About 15 hours before the outage, Teams traffic failed over from an alternative gateway to the primary gateway that also serves Exchange Online. The third was capacity: nearly 14% of gateway server capacity had been out of service since the previous day because of an unrelated issue. Microsoft said that shortfall alone caused no impact, but the redirected traffic then set off a cascade from server to server. At the lowest point, more than 80% of the servers had stopped responding, and that happened in under two hours.
The timeline shows a slow build. The first anomaly was detected at 4:03 p.m. UTC on Nov. 12, and infrastructure restarts began around 6 p.m. By 8:30 p.m. engineers found many servers unresponsive. A mitigation reverting the defective change on part of the capacity was applied at 7:20 a.m. UTC on Nov. 13. Service began stabilizing at 10 a.m., Asia Pacific recovered at 11 a.m., Europe was healthy by 11:35 a.m., and global webmail returned to expected levels at 2:17 p.m.
The recovery itself proved difficult. Exchange Online cannot fail out of this gateway for scale reasons, while Teams, OneDrive, SharePoint and Office apps could. The normal rollback, staged flighting and feature-flag tools could not revert the change quickly. Servers brought back into service were swamped by waiting traffic, and recovery batches covering 50% or more of a region had to be returned gradually. The safeguards meant to wall off partitions of the gateway did not stop the spread.
Accounts of the outage’s length differ. One summary of Microsoft’s analysis describes disruption of “more than twelve hours,” while the timestamps in the preliminary report span about 24 hours. Microsoft’s final post-incident report is due by Nov. 20.
The gateway is built from thousands of capacity units organized into regions, and Microsoft runs two independent, heterogeneous gateway systems. The repair items in the preliminary report carry a Nov. 20 completion date.
Sources: petri.com, office-watch.com.









