Was Microsoft 365 email down or not?

Microsoft 365’s mail service had troubles on Monday 4^th for almost 24 hours.  Or did it?  Microsoft’s reporting of the incident has been inconsistent and had more than the usual self-serving blather.

The Microsoft 365 email outage started on Monday 4^th September about noon UTC (early morning in the US) and was described by Microsoft as:

“Some users may be intermittently unable to access their Exchange Online mailboxes via any connection method”

“any connection method” is important because often Exchange Online outages are limited to one or two connection methods. For example, Outlook software can’t connect users can still use the web browser version.  On this occasion, all email access was disrupted.

According to Microsoft, the outage was fixed about 21 hours later (9:22am UTC on Tuesday 5^th).

The company also says that

“Impact may have occurred for some internal Microsoft users only.”

However the reports from DownDetector indicate this was more than an ‘in-house’ bug.

None of this was reported by Microsoft at the time.  We received three conflicting “365 Service Alert” emails many hours AFTER the outage was supposedly fixed?

What happened – was there an outage at all?

Microsoft’s story about this outage is more confused than usual. The first (belated) reports of the outage said that the cause was:

“ A code change made in a section of infrastructure which facilitated mailbox access caused the infrastructure to perform below acceptable performance thresholds, resulting in impact.”

The code change was ‘manually reverted’ and the problem allegedly resolved.

That report said “This is the final update for the event.”

But it wasn’t the last word.  Only 20 minutes later, another message arrives with a wholly different story (our text in bold)

“False Positive …
The investigation is complete and we’ve determined the service is healthy. A service incident did not actually occur. This communication will expire immediately.
This is the final update for the event.”

“Nothing to see here … move along”

We’ve seen Microsoft go into a state of corporate denial about problems and bugs but this is a new low. 

If the outage (sorry ‘service incident) didn’t happen, what about all the reports from users and services like DownDetector?  Were all Microsoft’s paying customers suffering from some mass delusion?

And how about Microsoft’s own admission of a code bug that had to be reversed?

The mystery deepens because Microsoft has deleted the issue completely. EX673494 (the source of the above quotes) doesn’t show up in ‘Issue History’ for Microsoft 365 Admins.

What to learn from this?

Bug and outage reports from Microsoft can’t be taken at face value. Microsoft’s corporate interests will always get in the way.

Microsoft will try to minimize the problem so it doesn’t look as bad as it really is. They have a whole glossary of weasel words and phrases to fall back on.

For example, you’ll often read that “Some users are affected …”. The word ‘some’ could mean anything from under 1% of customers to 99% .

There was nothing in Microsoft’s report for this or other outages about how customers restore their service. 

In theory, Outlook software should just reconnect when Microsoft’s servers are working but often it’s necessary to restart the software. That’s the kind of advice that Microsoft omits, probably because it doesn’t fit their ‘nothing to see here’ narrative.