A bug that had been sitting inside Microsoft's systems since July took Outlook, Teams, Copilot and other Microsoft 365 services offline around the world for nearly 24 hours this week, and the company now says it took three separate problems arriving at once to set it off

2

3

.

Source: Wccftech

Source: Wccftech

Three failures, one cascade

Per Microsoft's preliminary Post-Incident Report, dated November 15, the outage ran from 4:20 PM UTC on November 12 until 4:10 PM UTC the following day

2

1

. It names three contributing factors.

One was a latent defect introduced by a July Windows update on the servers that handle network traffic for Microsoft 365. Under certain workloads, it caused lock contention that exhausted CPU and memory. Another was routine: Teams traffic was automatically rerouted onto the primary gateway shared with Exchange Online about 15 hours before the outage began

2

3

. Microsoft typically runs two such gateway systems, and one was already not working for Teams

1

. The remaining factor was bad luck: nearly 14% of the gateway's capacity was already offline from an unrelated problem the day before

2

3

.

More than 80% of the gateway's servers had stopped responding within two hours, and automated traffic redirection overwhelmed the remaining ones in turn, per Microsoft's own executive vice president

2

3

1

. Regional containment safeguards failed to stop the spread

2

3

.

The bug escaped testing for more than three months, Microsoft says, because its labs ran on lower-core-count hardware, where the lock contention was less likely to appear

2

3

.

Why email was hit hardest, and why the fix took so long

Exchange Online suffered the greatest disruption, per the preliminary report, with many users unable to reach mailboxes or send and receive email

2

3

. Teams, Copilot, SharePoint and OneDrive had trouble with features that depend on Exchange Online

2

3

. Outlook and Exchange stayed tied to the failing gateway through peak US business hours while other apps shifted away

2

3

.

Source: Engadget

Source: Engadget

Recovery was slow for a structural reason: the bug was months old, so the standard rollback tools and feature flags engineers would normally use were useless

2

3

1

. Servers crashed again after restarts under accumulated traffic

2

1

. Reverting the code meant pulling over half the infrastructure in a region offline at once

2

3

1

. Investing.com reports it took roughly 15 hours to find the root cause and another 15 to begin rolling out the fix

3

.

The affected connection-handling code has been fixed, and webmail availability was back to normal worldwide as of 2:17 PM UTC on November 13

2

. The outage hit businesses, hospitals, law enforcement and governments

3

1

. Investing.com says institutional clients and market observers are likely to scrutinize Microsoft's load testing and failover safeguards

3

. It frames the episode as a reminder of the concentration risk in relying on a single cloud vendor

3

.

The report also says there is no sign of malicious activity or of AI agents playing a part

2

. The defect, it adds, is specific to Microsoft's own gateway infrastructure rather than Windows more broadly

2

. Charles Lamanna, Microsoft's executive vice president for Copilot and AI agents, apologized on X: "We are deeply sorry for the impact to all our customers and recognize the impact that this has on their normal business & operations"

1

.

Microsoft has since added traffic-distribution safeguards, expanded monitoring and increased capacity

2

1

. The queued email that backed up during the incident was later confirmed processed

2

. The company plans to publish its final report on the outage by November 20, with additional corrective measures to come in it

2

1

.

© 2026 TheOutpost.AI All rights reserved