Microsoft blames failed server health checks as recovery gains ground

Microsoft 365 went down around the world for roughly a day, taking Outlook email, Teams and the Copilot assistant with it. The outage began early Thursday afternoon Eastern time on November 12 and stretched into a second day, disrupting police forces, courts, hospitals and banks across several continents 2

2

1

1

.

The suite is the plumbing for a huge share of office work: Microsoft said in January it had more than 450 million paid commercial subscriptions 2

2

1

1

6

6

. Per Downdetector, roughly 130,000 user reports came in at the Thursday peak before waning by Friday morning 2

2

1

1

8

8

. One reading put the disruption across 74 countries and territories 2

2

. Microsoft did not say how many people were actually hit 2

2

1

1

.

A fix that didn't fix

Microsoft said there was no evidence of a cyberattack 1

1

7

7

3

3

4

4

. It blamed a problem during an earlier upgrade to network capacity that came to light after a recent network change 7

7

3

3

4

4

. For about a day the company offered no definitive technical explanation; its early updates described servers failing automated health checks and buckling under load 1

1

8

8

.

Early recovery efforts struggled. Some servers brought back online became unresponsive again once real traffic hit them 1

1

6

6

5

5

. Across 22 posts on X, Microsoft admitted its actions had not provided the relief it anticipated 1

1

6

6

5

5

. To keep delayed mail from bouncing, it extended its email timeout to 48 hours 1

1

.

Australia's Home Affairs Minister Tony Burke said Microsoft had likened the episode to the 2024 CrowdStrike failure, in which a faulty update grounded flights and disrupted hospitals and banks worldwide 2

2

. "They believe that they are dealing in some ways (with something) similar to what we saw with respect to CrowdStrike where there is an update that has gone awry," he said 2

2

.

Police, courts and banks

In the UK, police could not take online crime reports, a Manchester trial was halted with jurors sent home, and National Health Service email problems threatened prescription delays 2

2

1

1

6

6

. Norway's ministries were hit with no timeline for a fix, Japan's government IT infrastructure was affected, Singapore's government emails went down, and the Philippines Supreme Court's operations were disrupted 2

2

6

6

5

5

.

In the US, Baltimore County, Jacksonville and Austin had email problems 1

1

8

8

7

7

. Congressional staffers said policy negotiations were derailed. A Colorado sheriff's office said the outage made posting bond for prisoners' release difficult 1

1

8

8

7

7

. Workhorse Group delayed its quarterly earnings call. Danny Lau, who owns a Hong Kong aluminum firm, said his company could not get updates from US customers and feared for production timelines 2

2

1

1

6

6

. Some Hong Kong banks saw operations disrupted, and people were blocked from bank transfers that needed verification codes 2

2

1

1

6

6

. OpenAI and Anthropic each reported problems connecting with Microsoft 365 tools 2

2

8

8

6

6

. Air India rerouted its email and IndiGo's customer support suffered, though both airlines said flights were unaffected 6

6

.

The apology

By Friday Microsoft reported recovery progress, with email queues approaching normal levels, though some users still faced intermittent access and delayed mail 2

2

1

1

6

6

. "We are deeply sorry for the disruption this incident has caused our customers and partners," the company said 3

3

. It said the gradual pace was deliberate: engineers restored services in stages, validating stability at each step, which took longer but helped ensure a stable recovery 7

7

4

4

.

Microsoft said it had addressed the underlying issue and was seeing broad recovery and stability, though its own status page still showed a service degradation as of Friday evening 7

7

3

3

4

4

. Experts called the event historically large. Katie Moussouris, a security researcher who once worked at Microsoft, said "It's the global scale of it and the duration that make it historic" 7

7

. Ciaran Martin, a former head of the UK's cyber security agency, called it possibly among the biggest outages ever recorded 1

1

7

7

.

The outage also landed amid questions about Microsoft's own cost-cutting. The company had offered buyouts to about 60,000 employees last month, more than a quarter of its workforce, after laying off 9,000 the year before and 10,000 in 2023 7

7

3

3

. Some experts and users suggested the cuts played a part, without offering evidence 7

7

3

3

.

The episode fed an argument Western European governments were already having about relying on a few big US tech providers 6

6

5

5

. Germany's digitization ministry said German business was widely affected but government disruption was minor, because the federal government does not use Microsoft in the cloud 6

6

5

5

. Greece's digital governance minister said affected ministries could still work and urged building technological alternatives to Microsoft's services 6

6

5

5

. Commentators said the outage shows the risk of depending on Microsoft's digital infrastructure. University of Glasgow computing professor Jeremy Singer put the mood among Microsoft's customers more plainly: "With no emails coming or going, many office workers feel like they are having a virtual snow day" 6

6

.