Microsoft’s Final Report Names the July Bug That Broke 365

Human Verified

Share to: , and

A dark server hall with amber warning lights glowing along the racks

Microsoft has published the final report on the November outage that knocked out Exchange Online and Microsoft 365 for almost a full day. The report names four causes, led by a defect hiding in gateway servers since July. All repairs are complete, and the remaining fixes run into 2027.

What the Final Report Adds

The final report replaces the preliminary version Microsoft issued on November 15, and it changes the picture in one key respect. That report named three factors behind the outage; the final report names four, adding a Teams failover that shifted traffic onto the Exchange gateway hours before the collapse.

Microsoft says the outage began at 16:20 UTC on November 12 and active impact ended at 16:10 UTC on November 13, with the incident declared resolved on November 14. Exchange Online was hit hardest, while Teams, SharePoint and OneDrive recovered earlier after their traffic moved to a redundant gateway.

No mail was lost, Microsoft says, and queued messages were delivered after the retry window was extended from 24 hours to 48. The report states the outage was not a cyberattack, no malicious activity was found, and the change behind the defect was not made by AI tooling.

Four Factors, One Cascade

The first factor is a latent defect in connection processing inside an internal Windows process, where tasks waited on each other and exhausted CPU and memory. It came from an update rolled out to gateway servers from July, and it struck only newer, high-core servers that load testing had missed.

The other factors were reduced capacity, with nearly 14 percent of gateway servers out of service for routine work, and a Teams failover at 01:07 UTC that moved Teams traffic onto the Exchange gateway. The fourth was peak load, as failures began when United States traffic approached its daily high.

Together they exposed the defect, and the cascade was fast. Within two hours, more than 80 percent of North American gateway servers were unresponsive, and traffic sent to other regions met the same defect there. The revert update reached part of the fleet at 07:20 UTC on November 13.

Our Read

The report is unusually candid about a failure months in the making, and the July detail should worry other operators most. A change can sit quietly in production for three months, pass every test, and still bring down a service once the right mix of load, capacity and failover arrives.

Microsoft’s repair list is complete, but the next steps are the real test. The company plans smaller fault zones, workload-isolated gateway pools and load testing on its newest hardware, with items due by the first quarter of 2027. The question is whether those changes land before the next defect does.

Why It Matters

One of the year’s biggest cloud outages came not from an attack or a bad deploy, but from a months-old defect meeting ordinary conditions. Microsoft’s final report shows how that combination turns a normal morning into a global outage — a warning for every operator running a shared gateway.

Background

The outage hit Exchange Online, Copilot, Teams, SharePoint, OneDrive, Defender and eight other services across all of Microsoft’s global fleet. Service was restored on November 14 after extended monitoring, and the final report, dated November 20, replaces the preliminary version entirely.

Source: Microsoft’s final report

Share to: , and

Faheem R. Khattak

Senior automation engineer and AI marketing journo covering LLMs, algorithms, analytics, core updates, Bing search, and technical SEO.

0 Comments

Leave a Reply

Optional Take Tag:

🛡️ Spam protection: Honeypot + Time-Trap Active

This site uses Akismet to reduce spam. Learn how your comment data is processed.