Microsoft's preliminary Post-Incident Report attributes a near 24-hour Microsoft 365 and Exchange outage to three factors that combined. A change deployed to the cloud gateway in July created a latent bug that could exhaust CPU and memory under certain conditions. Microsoft says the defect was missed in pre-release testing because the test hardware was less powerful than some production systems.

The report says traffic shifts exposed the bug. About 15 hours before the outage, Microsoft moved Teams traffic from the alternate gateway to the primary gateway, which handles Exchange Online access. Microsoft describes the switch as a standard fail-safe, but the primary gateway then handled more traffic with a different request mix.

Microsoft also reports that its gateway server fleet was operating 14 percent below optimal conditions due to an existing fault. According to the report, that lost capacity left too little headroom to absorb the changed demand.

The report states the outage ran from November 12 at 16:20 UTC to November 13 at 16:10 UTC. As US business hours peaked, more than 80 percent of affected gateway servers froze. Backup gateways showed the same lockups. Microsoft's report does not explain why regional isolation mechanisms failed.

Microsoft says recovery required a targeted system update from its engineers. Restarts failed because queued requests triggered resource exhaustion again when servers returned. Rollbacks also failed because the July defect had not been identified. Microsoft says Teams, OneDrive, and other Office applications could be rerouted away from affected infrastructure, but scaling limits prevented Exchange Online from doing the same. Microsoft extended delivery timeouts so delayed email would not bounce as undeliverable.

Microsoft says it has added capacity and traffic-distribution guardrails. It plans to publish its full final incident report on November 20. The preliminary report is covered by incident IDs EX1550543 and MO1550821.