Microsoft Traces Exchange Online Outage to a Months-Old Windows Fix on Its Gateway Servers
Microsoft's preliminary incident report blames the 12 November Exchange Online outage on a latent Windows defect in its gateway servers, made worse by a Teams failover and missing capacity. Delayed mail was held, not bounced.

Microsoft has explained what broke Exchange Online for almost a full day. In its preliminary Post Incident Report for incident EX1550543, dated 15 November 2026 and published to administrators in the Microsoft 365 admin center, the company traces the 12 November outage to a months-old Windows fix on the networking gateway servers that carry Microsoft 365 connections. An abridged version was posted publicly by Microsoft executive Charles Lamanna, who apologised for the impact on customers.
The report confirms what senders saw on the wire: mailboxes were unreachable, inbound mail queued for hours, and Microsoft stretched its delivery window so that delayed messages would not bounce back as undeliverable. The final report, with the full list of repair items, is due within five business days of full resolution, which points to around 20 November.
Key facts
- Incident: EX1550543 (Exchange Online) and MO1550821 (Microsoft 365)
- Impact window: 12 November 2026, 16:20 UTC, to 13 November 2026, 16:10 UTC, about 23 hours 50 minutes
- Root cause: a latent defect in a Windows fix deployed to Microsoft 365 networking gateway servers since July
- Triggers: extra load after Teams failed over to the gateway shared with Exchange Online, and about 14% of gateway capacity already out of service
- Worst point: more than 80% of gateway servers unresponsive
- Mail handling: delivery window extended so delayed messages were not returned as undeliverable
- Next: final Post Incident Report expected around 20 November 2026
What Microsoft says went wrong
According to the preliminary report, three conditions lined up. The first was a defect inside a Windows fix built specifically for the gateway servers in front of Microsoft 365. It had been rolling out since July and ran for more than three months without trouble, but under certain conditions it caused lock contention inside an internal Windows process, which exhausted CPU and memory on the server. Microsoft says neither the defect nor the fix applies to general Windows deployments, so customer machines were never part of the problem.
The second condition was load. About 15 hours before the outage, Teams traffic failed over to the primary gateway that also serves Exchange Online. The third was capacity: nearly 14% of the gateway servers were already out of service because of an unrelated issue the day before. Microsoft says the remaining capacity was sufficient on its own, but together the three factors pushed servers into the defect.
Once the first servers failed, their traffic moved to healthy ones, which then failed in turn. Microsoft describes "a cascading failure from server to server" and says that "in less than two hours, the previously unknown defect was encountered across our fleet." The bulkheads that normally separate capacity units did not contain the failure because of a separate issue, and at its lowest point more than 80% of servers were unresponsive.
Why Exchange Online took the hardest hit
Teams, OneDrive, SharePoint and the Office apps could be moved off the faulty gateway path. Exchange Online could not, for scale reasons, according to Microsoft. That is why Outlook, Outlook on the web and other mail clients lost access to mailboxes while some other services recovered sooner, and why mail flow was the part of the incident senders felt most.
Recovery was also slow for a structural reason. Because the defect was months old, the usual rollback and feature flag tools could not revert it quickly. Engineers built an update that returned connection handling to its earlier, known good state, then took large groups of servers out of service, applied the change and restarted them. Restarted servers were immediately flooded by waiting traffic, so traffic had to be brought back gradually.
The recovery timeline
The report gives this sequence, in UTC. On 12 November, a connectivity anomaly was detected at 16:03 and a high-priority alert raised at 16:20. On 13 November, Microsoft applied the mitigation to part of the capacity at 07:20, Asia Pacific recovered around 11:00 and Europe returned to healthy levels at 11:35. Mail queue backlogs started to shrink at 13:00, global webmail availability returned to expected levels at 14:17, and from 16:10 customers were no longer actively affected. Microsoft then monitored until all queued mail had been processed, and issued its final customer updates on 14 November at 04:15. Our live coverage of the outage tracked these stages as they happened.
What Microsoft has changed
Microsoft lists five actions as completed: connection handling code returned to a known good state, the email delivery window extended "to ensure delayed messages weren't returned as undeliverable", extra capacity added for recovery, more monitoring of traffic redirection and connection health, and new guardrails on traffic distribution to stop a failure cascading across the fleet. The full repair list will come with the final report, which Microsoft says supersedes the preliminary one.
What it means for senders
What changes
Microsoft has confirmed that the 12 November Exchange Online outage came from a defect on its own gateway servers, not from senders, DNS or authentication, and that it extended its delivery window so delayed mail would not bounce. Deferrals to Microsoft recipients during the window say nothing about your reputation.
Who is affected
Anyone who sent to Outlook, Hotmail and Microsoft 365 addresses between 12 November, 16:20 UTC, and 13 November, 16:10 UTC: ESPs, marketing teams, transactional senders and IT teams that run Exchange Online.
When
The preliminary report is dated 15 November 2026. The final Post Incident Report is expected around 20 November 2026.
Action checklist
- Exclude the outage window from deliverability reporting, so deferrals and late opens from 12 and 13 November do not distort your Microsoft metrics.
- Check for real bounces on messages sent to Microsoft domains during the window and resend the ones that still matter, such as password resets and invoices.
- Do not change SPF, DKIM, DMARC or IP setup because of this incident: Microsoft places the failure inside its own infrastructure.
- Review your retry settings so your MTA keeps retrying 4xx deferrals for long enough. Our guide to 421 Service Temporarily Unavailable explains how these deferrals work.
- Distrust outage-themed emails that ask for a click to recover mail: attackers are already using the incident number as a phishing lure.
- Read the final report in Service health in the Microsoft 365 admin center under EX1550543 when it is published.
Sources
- Microsoft, preliminary Post Incident Report for EX1550543, dated 15 November 2026, Microsoft 365 admin center, Service health (available to tenant administrators)
- Charles Lamanna, Microsoft, on X: abridged root cause and apology
- Microsoft 365 Service health status
Related on SMTPedia
SMTPedia News follows our editorial standards. Spotted an error? See how we correct it.
Get SMTPedia news first on Google
Add us as a preferred source to see our stories in Top Stories and Discover.
Add as preferred sourceUseful story?
SMTPedia news is free and independent. A coffee helps us keep covering what changes for senders.
Buy SMTPedia a coffeeNewsletter