For the better part of a day last week, most people who relied on Exchange Online could not reach their mailboxes, and Teams, SharePoint, OneDrive and Copilot faltered alongside them. Microsoft has now said the trigger was a defect that had been sitting inside its Microsoft 365 networking gateway since July, undetected for more than three months.
The company laid out the explanation in a preliminary post-incident report dated Nov. 15, covering an outage that began Nov. 12 and ended Nov. 13. A final report is still to come.
A July update that ran quietly for three months
Microsoft’s preliminary post-incident report lists incident identifiers EX1550543 and MO1550821 and puts the start at 4:20 p.m. UTC on Thursday, Nov. 12 and the end at 4:10 p.m. UTC on Friday, Nov. 13. It says the outage was “caused by three main factors that together caused a global outage for the networking gateway.” The first was the defect itself: “a latent defect caused Windows servers to become overwhelmed and unresponsive under certain load conditions.”
The change that carried the defect “first started deployment in July of this year,” the report says. It rolled out globally over several weeks and ran for more than three months with no observed impact. Ordinary traffic never pushed the gateway servers into the conditions that expose the flaw.
Two pressures that woke the defect
The other two factors were circumstantial. About fifteen hours before the outage, according to the report, Teams failed over to the primary gateway that also serves Exchange Online, raising load and changing traffic patterns. And nearly 14% of gateway server capacity was already out of service because of an unrelated problem the previous day, so the same traffic landed on fewer machines.
BleepingComputer’s account of the report adds that Microsoft’s load testing used less powerful hardware than the machines in production, so the lab traffic mix never exposed the bug. Petri described the same gap this way: validation environments leaned on older servers, while the defect was more likely to appear on newer, higher-core-count ones under particular workloads.
How failover spread the failure worldwide
Once the defect tripped, the gateway’s own safety systems made things worse. Automated redirection moved traffic from failing servers to healthy ones, which carried the same bug and failed in turn. Microsoft’s report puts the cascade at less than two hours from first encounter to fleet-wide failure, and at the worst point more than 80% of servers were unresponsive. Regional safeguards designed to contain such a spread did not hold, according to BleepingComputer’s reading of the report, so a fault that began in one place became a global one.
Exchange Online had the most significant impact, and the report says most users could not reach their mailboxes. Unlike Teams, OneDrive and SharePoint, Exchange could not shift to alternative infrastructure. The services affected, per the report, ran far beyond email: Copilot, Teams, SharePoint, OneDrive, Defender, Purview, Fabric, Intune, Planner, Power Apps, To Do, Power BI and Universal Print.
Recovery was slow for a reason that is easy to miss. Because the faulty code had been deployed months earlier, Petri reports, a simple rollback was not possible. Restarted servers were quickly swamped again by the backlog of waiting traffic. Engineers instead built a targeted fix, took groups of servers offline, applied it and brought capacity back gradually; the report says fixes went to groups of 50% or more of a region at a time.
The public timeline on the day shows how long the cause stayed hidden. As BleepingComputer tracked it live, Microsoft first pointed to elevated infrastructure load, then said in an X post that the load was not driven by an increase in traffic. Its early response was to pause deployments, redistribute traffic and restart affected servers. By 6:57 p.m. ET on Nov. 12 it reported limited recovery and said a core process on affected servers had become overwhelmed. By 8:05 a.m. ET on Nov. 13 it had extended the email timeout window to 48 hours so that delayed messages would be preserved, and by midday email queues had dropped significantly. A separate Teams incident earlier on Nov. 12 ran from 9:45 to 10:53 a.m. ET, and Microsoft said it was unclear whether that incident was connected.
What Microsoft says it has changed
While the outage was still being worked, Microsoft said on Nov. 13 that it was “a technical issue not related to cyber or AI / agents,” a statement the security newsletter from the IT firm Cadre repeated in its Nov. 17 edition alongside the finding that the cause was a bug dormant since July. Charles Lamanna, Microsoft’s executive vice president for Copilot, Agents and Platform, apologized for the disruption in a post on X, BleepingComputer reported, and The Stack quoted his description of the July change as a fix “prepared specifically for Microsoft 365 networking gateway servers.”
The report lists completed repairs: the connection-handling code path was reverted to a known good state, the email delivery window was extended so delayed messages were not returned as undeliverable, extra capacity was added, monitoring was put in place for traffic-redirection patterns and connection-handling health, and guardrails now limit how redistributed traffic can cascade. What remains is the final report. The repair table in the preliminary document gives its due date as Nov. 20, 2026, within five business days of full resolution, though the body text says “later this week.” That document is where Microsoft has promised a full root cause analysis and longer-term repair items, including why the regional safeguards failed to contain the cascade.
This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.
More from Morning Overview
- Cadillac cut the price of its cheapest EV, which goes up to 317 miles on a charge
- Winter weather advisories cover I-80, I-70 and the Bighorns with up to 15 inches of snow
- Arizona’s governor declared a statewide emergency after Flagstaff logged its wettest Nov. 1 on record
- Zero Motorcycles unveiled an electric motorcycle targeted at under $8K and a street-legal dirt bike at EICMA