Outlook went down for four hours Wednesday. What happened?

The version of Outlook that runs on Windows PCs suddenly stopped working worldwide on Wednesday. The outage wasn’t caused by a bad patch. It wasn’t restored by an update. Microsoft isn’t saying much, but there’s a plausible explanation - and it’s disconcerting.

Email migration to Microsoft Outlook app on mobile email for smartphone user.
Suwaree Tangbovornpichet / Getty Images

At mid-morning Wednesday in the U.S., Outlook stopped working. Dead in its tracks. Both the Click-to-Run version of Office (er, Microsoft) 365 and the installed (MSI) version of Outlook refused to start

Today when I went to check my email, Outlook would not open; it would load the “Starting Outlook…” splash screen, which would close without opening the Outlook window itself, and the taskbar icon went away. Looking in Reliability History, it states that Outlook has crashed.

I tried opening in safe mode (it does the exact same process as described above) and restarting the computer to no avail. I even tried the full repair (not the quick one), redownloading and reactivating MS Office 2019, but no go.

Then, four or so hours later, Outlook suddenly started working again.

There are so many programs running around called “Outlook” that it’s difficult to nail down which versions of Outlook went AWOL. But it looks like all of the recent versions of Microsoft 365 Click-to-Run Current Channel – dating back to June 30’s Version 2006 Build 13001.20266, at least – as well as Office 2019 packed their bags and left the station.

Microsoft, in its inimitable way, left a trail in its @MSFT365Status tweets:

We’re investigating an issue affecting user access to Outlook. We're investigating whether a recently deployed update could be the source of this issue. As a workaround, users can utilize Outlook on the web or their mobile clients… (four hours later) We're rolling out a fix for this issue, and we expect the mitigation to reach all customers over the next few hours. 

So for four hours yesterday, you could get at your Outlook email through your iPad, or by hitting the website, but the version of Outlook you probably use on your PC had taken the morning off.

Ends up the sudden disappearance was attributable to the Outlook.exe program crashing with an error 0xc0000005. Lawrence Abrams at BleepingComputer has full details.

Two obvious questions present themselves:

  1. What broke? There weren’t any new versions of Office pushed out yesterday. Yet, all over the world, millions (if not hundreds of millions) of Outlook users all had the rug pulled out from under them.
  2. How did Microsoft fix it? There was no patch delivered, no new C2R version, no emergency update for Outlook 2019. Microsoft’s resorted to some odd patch delivery mechanisms lately (notably using the Microsoft Store) and it’s had multiple-cumulative-update problems. But this fix apparently didn’t touch PCs at all.

Microsoft’s proffered explanation doesn’t pass the sniff test:

There is a new symptom of Outlook crashing on launch starting on 7/15/2020.   A fix has been published but will take time to propagate to worldwide availability. Outlook will automatically look for the fix on launch, so if this issue persists through multiple launches please use Outlook Web Access (or your providers webmail interface) for an hour then try again. This problem is not associated with any of the 7/15/2020 security patches so there is no need to uninstall them if Outlook will not launch. 

In fact, there weren’t any changes made to any PCs. Individual users didn’t get new versions. Administrators didn’t push anything out on their networks. Outlook just suddenly started working again.

Many people discovered that they could get Outlook working by rolling back a few versions. Older versions of Outlook (both C2R and MSI) worked. The newer ones didn’t. The bug had nothing to do with specific patches.

I was contemplating the mysteries of life, the universe and Outlook appcrashes, when an anonymous post appeared on AskWoody:

Woody – can’t tell you my source but basically rumor is something was pushed server-side on Microsoft’s end in regards to a security or authentication change (who knows what specifically) and in that change it cut off communication with Outlook clients past a certain patch. I confirmed this was not last night’s patch that contained these changes, but one previously issued. I do not know anything more specific about this patch in question. This is why rolling back worked – it went to a client version that was before that change in how it talked to their servers. Why this was done in the middle of the day, I don’t know. Very unusual. All I’m willing to go into right now.

That explains it. (Thank heaven for anonymous posters!) As @NetDef says:

Admins: do not need to check for updates or re-apply updates. Just start Outlook. If you rolled back earlier today it’s safe to roll forward again to bring you back to current patched status.

The bug was a combination of a new security check on the client, and a back-end server patch/change. For now the back-end server change has been reverted for Office 365 Exchange tenants.

Translating that into English: Microsoft built some fancy new checking mechanism into the more recent versions of Windows-based Outlook. It was working just fine until, yesterday morning U.S. time, somebody changed something on Microsoft’s servers. Kaboom. No more Outlook.

Microsoft claims:

Root cause: A recent change to the Outlook client inadvertently resulted in this issue.

That’s sorta true. What clobbered everybody was some stupid change on a server.

Next steps:

– We’re reviewing our update procedures to isolate these issues prior to our deployment cycle.

– We’re analyzing our telemetry data to better identify these issues before they impact our customers. We’ll publish a post-incident report within five business days.

I could ask the obvious questions – most beginning with, “What in the blue blazes…?” or something more colorful – but you can fill those out yourself. Bottom line: Somebody at Microsoft changed something on a server, without testing the change, and it brought down Windows-based Outlook everywhere.

Deployment "cycle"? Puh-leeze. We saw the same stupidity with the Windows Search box bug back in February. Somebody threw a monkey wrench (er, spanner) in the server. Everybody got clobbered.

Kinda makes you feel warm and fuzzy.

If you thought an installed Windows version of Outlook would be more stable than a cloud version – or an iPad or Android or iPhone version, for that matter – this is what’s known as a comeuppance. 

We’re always looking for answers on AskWoody.

Copyright © 2020 IDG Communications, Inc.

  
Shop Tech Products at Amazon