GBHackers

Microsoft Reveals Windows Bug That Crashed Over 80% of Servers During Global Microsoft 365 Outage


Microsoft has attributed the global Microsoft 365 disruption on November 12–13, 2026, to a latent Windows defect that rendered more than 80% of its affected networking gateway servers unresponsive. Its preliminary incident report, dated November 15, describes how lock contention, redistributed traffic, and reduced capacity combined to trigger a cascading infrastructure failure.

The Microsoft 365 outage primarily disrupted access to Exchange Online mailboxes and delayed email delivery. Active customer impact lasted from 4:20 PM UTC on November 12 until 4:10 PM UTC on November 13, totaling 23 hours and 50 minutes. Other applications experienced varying disruption where their features depended on Exchange Online.

Windows Bug Crashed Over 80% of Servers

According to the preliminary report, the defective code originated in a fix prepared specifically for Microsoft 365 networking gateway servers. Deployment began in July and expanded globally over several weeks without observable problems. Microsoft emphasized that neither the underlying defect nor this specialized fix applies to broader Windows deployments.

Under particular traffic conditions, lock contention inside an internal Windows process exhausted CPU and memory resources, preventing servers from handling requests. Testing missed the failure because laboratory traffic and hardware differed from production. Notably, newer machines with more processor cores were more susceptible, while testing emphasized weaker hardware.

About 15 hours before the outage, Teams automatically moved from an alternative gateway to the primary infrastructure supporting Exchange Online. Unlike earlier Teams meeting disruptions, this event involved shared networking infrastructure rather than isolated collaboration functionality. The routine failover changed traffic patterns and increased pressure on the primary gateway.

Meanwhile, an unrelated issue had temporarily removed nearly 14% of gateway capacity. Microsoft said sufficient headroom remained, meaning that reduction alone would not have caused disruption. Combined with shifted traffic, however, it exposed the dormant contention bug during peak demand in the United States.

As servers stopped responding, automated traffic management redirected connections toward healthy endpoints. Those machines then hit the same defect, creating a feedback loop as capacity shrank. Within less than two hours, over 80% of gateway servers were unresponsive. Separate isolation failures allowed the cascade to cross infrastructure partitions globally.

Teams, SharePoint, OneDrive, and Office applications could redirect eligible traffic away from the faulty gateway. Exchange Online could not do so at the required scale. Administrators tracking EX1550543 and MO1550821 can review tenant advisories through Microsoft’s Service health dashboard, which distinguishes active incidents from resolved service issues.

Repeated restarts failed because returning machines immediately absorbed waiting traffic and became unresponsive again. Engineers ultimately developed an update that restored the previously working connection-handling code, removed server groups from service, applied the change, restarted systems, and gradually reintroduced traffic. Mitigation began reaching affected capacity at 7:20 AM UTC on November 13.

Microsoft also expanded capacity, extended email delivery windows, strengthened connection monitoring, and added traffic distribution guardrails. The preliminary report describes an operational defect, not attacker exploitation. Separately, outage-themed phishing illustrates how criminals can exploit disruption through misleading recovery messages, creating additional risks while organizations restore normal operations.

A final incident report will supersede these preliminary findings and detail additional corrective measures. Microsoft listed November 20 for its comprehensive repair plan, leaving longer-term testing and isolation improvements pending further disclosure.

Give your SOC the alert context on the latest threats to stop incidents quickly -> Integrate Threat Intel from 16K SOCs:



Source link