Technology

Add as a preferred source on Google

Microsoft Says More Than 80% of Gateway Servers Went Unresponsive

A Teams failover, a software defect and reduced capacity combined to disrupt Exchange Online for about a day, according to Microsoft's preliminary review.

Updated 3 min read
An employee checks a phone outside Microsoft’s campus in Redmond, Washington. / TokenPost.ai (mono)
An employee checks a phone outside Microsoft’s campus in Redmond, Washington. / TokenPost.ai (mono)

Microsoft said a latent gateway defect, a Teams failover and reduced server capacity combined to disrupt Exchange Online for about a day, showing how shared cloud systems can carry failures across services.

The company recorded active impact from 11:20 a.m. ET Thursday, Nov. 12 (16:20 UTC) through 11:10 a.m. ET Friday, Nov. 13 (16:10 UTC). Most people trying to use Exchange Online could not reach their mailboxes, and many experienced delayed email delivery.

The sequence began earlier: Teams traffic moved from an alternate gateway to the primary gateway shared with Exchange Online at 8:07 p.m. ET Wednesday, Nov. 11 (01:07 UTC Nov. 12), roughly 15 hours before Microsoft recorded the Exchange impact. Nearly 14% of that gateway's capacity was unavailable because of a separate issue from the previous day, Microsoft said.

A fix to Microsoft 365 networking gateway software contained a latent defect. Microsoft began deploying it in July and had not detected trouble for more than three months. The company's testing setup did not reproduce production conditions, Microsoft said; the defect was more pronounced on newer servers with more processor cores.

When servers stopped responding, automated traffic management sent requests to the remaining healthy machines, increasing their load. At the outage's lowest point, over four-fifths of gateway servers were unable to respond, Microsoft said. Some servers failed again as Microsoft returned them to service and traffic came back, slowing recovery.

Microsoft said its automated recovery system could not restore the gateway from the widespread unresponsive state. Separately, the tools available to roll back changes could not quickly undo the latent defect; Microsoft reverted the affected code path and brought service back in stages.

Microsoft operates two separate gateway systems for isolation and availability. Teams, OneDrive and SharePoint were among services that could shift away from the affected gateway, while Exchange Online could not move traffic on the same scale. Microsoft said most other affected services that relied on Exchange Online stayed largely online and accessible.

The Associated Press reported that Downdetector logged outage reports from 74 countries and territories within about an hour. The AP also reported that authorities in Japan, Singapore and Australia said some operations were affected.

Microsoft's preliminary report gives no total count of affected users. The company said in a December 2025 blog post that more than 430 million people used Microsoft 365 apps, but that figure is not a count for this incident. Jeremy Singer, a professor at the University of Glasgow's School of Computing Science, told the AP the outage highlighted “the dependence we have built on one company's digital infrastructure.”

Microsoft now says a final report with a comprehensive repair list is scheduled for Friday, Nov. 20.

Related articles

View all