When Outlook and Microsoft 365 went down last week, most of us lost a day of email and a few Teams calls. At a theme park hotel near Cologne, Germany, the outage had a much more physical cost. A pool safety system spotted a dangerous problem and sent an alarm by email. Nobody received it, and 15 guests ended up with breathing problems.
The incident happened at the Hotel Ling Bao, part of the Phantasialand theme park in Brühl. German broadcaster WDR first reported it, and the park confirmed the details to the broadcaster. As UNILAD Tech put it, it’s a worrying example of what happens in the real world when the systems we quietly depend on fail.
The short version
- The automatic dosing system for the pool in the hotel’s sauna area malfunctioned and pumped in too much chlorine.
- The maintenance and safety system raised the alarm, but only by email.
- Because of the Microsoft outage, the emails never arrived, Phantasialand told WDR.
- 15 guests reported respiratory irritation. 4 went to hospital, and 1 stayed for observation with a moderately serious injury.
- The pool was closed, the water was replaced, and it has since reopened.
What happened at Hotel Ling Bao
The trouble started on Friday, November 13, while Microsoft was still struggling to bring Exchange Online back. Guests who had used the pool in the hotel’s sauna area began complaining of health problems. Emergency services and the fire department were called.
At first, 13 guests reported irritation of their airways. That number later rose to 15. Early reports said five people were taken to hospital, which was then corrected to four. One of them remained under observation because of a moderately serious injury.
When responders tested the system, they found a defect in the pool’s automatic chlorine dosing system. It had pushed the chlorine level far above where it should be. The safety system did exactly what it was supposed to do and fired off warnings. The problem was where those warnings went.
How one outage turned into a safety incident
1. THE CLOUD FAILS
Microsoft 365 and Outlook go down on November 12 and stay patchy for about a day.
2. THE PUMP FAILS
The pool’s automatic dosing system develops a defect and adds far too much chlorine.
3. THE ALARM FIRES
The safety system detects the problem and sends a warning email to staff.
4. NOBODY SEES IT
The email doesn’t arrive. Guests keep using the pool until they start feeling sick.
15
guests affected
4
taken to hospital
1
kept for observation
0
alarm emails received
Why too much chlorine is dangerous
Chlorine is what keeps public pools safe from bacteria, so a pool without it would be a problem too. But the dose matters. German public pools follow the DIN 19643 standard, which typically calls for free chlorine of roughly 0.3 to 0.6 milligrams per liter in a regular pool. Dosing systems measure the water constantly and add more when levels drop.
When one of those systems keeps pumping, the water becomes harsh on skin and eyes, and irritating gases, including chlorine and chloramines, can build up in the air just above the surface. That’s especially true indoors, where there’s less airflow. Breathing that air leads to coughing, a burning throat, chest tightness and, in worse cases, lung irritation that needs medical treatment. For a sense of how little it takes, OSHA’s workplace exposure ceiling for chlorine gas in the US is just 1 part per million.
Signs a pool may be over-chlorinated
A strong, sharp chemical smell (a well-run pool barely smells), stinging eyes after a few minutes, coughing or a scratchy throat near the water, or skin that itches right after you get out. If several people notice it at once, leave the area and tell staff. Fresh air is the first step, and anyone with chest pain or trouble breathing should get medical help.
The bigger problem: email-only alarms
The pump failure was a mechanical fault. Those happen. What turned it into an incident was that the warning had only one way to reach a human, and it ran straight through a cloud service that was down.
That’s a surprisingly common setup. Building management systems, freezers in labs and restaurants, server rooms, water treatment and plenty of industrial equipment can be configured to send alerts by email because it’s cheap and easy. Most of the time it works fine. But email isn’t built for urgent alerts. It can be delayed, filtered as spam, or, as last week showed, simply not delivered for hours.
| Alert channel | Speed | Depends on the cloud? | Good for safety alarms? |
|---|---|---|---|
| Seconds to hours | Usually, yes | Not on its own | |
| SMS text | Seconds | Mobile network | Good backup |
| Automated phone call | Seconds | Phone network | Good backup |
| App push / paging service | Seconds | Yes, but separate from email | Good, with escalation |
| Local siren or warning light | Instant | No | Yes, for anything on site |
The fix isn’t complicated. Safety-critical alarms should use at least two channels that don’t share a single point of failure, and at least one of them should work even if the internet doesn’t. A flashing light and buzzer in the plant room, plus a text message, would very likely have caught this. Good alerting tools also escalate: if nobody acknowledges an alarm within a few minutes, it goes to the next person on the list.
A week of single points of failure
The timing is almost too neat. On Thursday, Microsoft’s outage knocked out Outlook, Teams and Copilot for millions of people. We covered what caused the Microsoft 365 outage and how businesses can claim credits, and the answer was a network upgrade that hid a problem until a later change set it off. On Friday, a router fault during a fiber upgrade led to the FAA’s four-hour nationwide ground stop, which disrupted about 15,000 US flights.
Different systems, same lesson. A huge amount of everyday life now runs through a handful of platforms, and when one of them stumbles, the effects show up in places nobody planned for. According to figures cited by UNILAD Tech, more than 430 million people use Microsoft 365 apps. Very few of them would have guessed that a hotel pool was one of the things relying on it.
Microsoft’s outage, by the numbers
Based on Microsoft status updates and Downdetector reports
Downdetector reports in the first ~30 minutes: 129,000+
Hours some users were locked out: 18+ (of a ~24-hour incident)
Email retry window Microsoft extended to: 48 hours (from 24)
That last number matters here. Microsoft extended how long it would keep retrying stuck messages, so many delayed emails were eventually delivered. For a normal inbox, that’s great. For an alarm about a chemical leak, an email that shows up hours late is about as useful as one that never arrives.
If you run any kind of alarm on email, check this today
- List every system that sends alerts by email: building controls, freezers, server rooms, security cameras, pumps.
- Add a second channel that doesn’t depend on your email provider, such as SMS, a phone call or a local alarm.
- Set up escalation so an unacknowledged alarm goes to someone else after a few minutes.
- Test it during an outage drill. Unplug the internet and see what still reaches a person.
- Keep a manual check. For anything involving chemicals or people’s safety, a scheduled human reading is a cheap safety net.
The takeaway
Nobody at Phantasialand ignored a warning. The warning just never got there. The pump failed, the sensor worked, and the message got stuck in the same outage that was frustrating office workers all over the world. Fortunately, every guest is accounted for and the pool is open again. But it’s a clear reminder that “we’ll get an email” isn’t a safety plan, and that any alarm that really matters needs more than one way to reach a human.
About this article: GeekBlog covers U.S. technology news, AI, phones, smartwatches and gaming. Every story is written and checked under our Editorial Policy. Spotted a mistake or have a story tip? Contact our editors.

