Tornado Datacenter experienced a notice incident on May 10, 2026 affecting Air Conditioning, lasting 4h 48m. The incident has been resolved; the full update timeline is below.
Affected components
Update timeline
- investigating May 10, 2026, 09:14 AM UTC
On-Site engineers investigating a current issue with aircon unit #2 appearing to be a electrical problem with the outside unit (condensor). We have temporarily increased supply temperature on other units to stabilize cold aisle temperature.
- identified May 10, 2026, 10:26 AM UTC
The unit is currently out of service due to a triggered security valve on the refrigeration circuit, likely causing impact to the refrigeration loop. We're cooling the datacenter with the working cooling units and have informed the maintenance company about the same.
- monitoring May 10, 2026, 12:26 PM UTC
On-Site engineers still monitoring the situation while the unit is out of service and being repaired.
- resolved May 10, 2026, 02:03 PM UTC
We will provide more details in a planned post-mortem. Temperatures throughout the data hall are stable since the incident happened in the morning, meaning redundancy works as intended. We observe a operational temperature range of about 25.4°C median across all cabinet cold aisle inlet sensors. On-Site engineers monitoring the situation until we have switched back thermal loads to free cooling later in the day. We will leave cold air supply increased, until repair operations are finished. For the upcoming week, we expect the pipe construction company to finish welding works on our new cold water loop network. The same supplies our in-house engineered cooling solution called deltaflow, offering advanced cooling output.
- postmortem May 11, 2026, 12:46 PM UTC
On 10th May 2026, 10:12 CEST, we experienced a failure of aircon unit #2 within RR.A. The issue was caused by a failure of the refrigeration loop due to a electrical issue with the outside condensor, causing blower fans to stop operating properly. The said issue caused increased pressure, which triggered a safety valve within the refrigeration loop, right after the compressor of the unit. Due to the triggered security valve, the unit was unable to cool properly anymore. Since the devices do not stop the inside EC fan, which is situated right after the vaporizer, the unit sucked warm air back into the cold aisle, causing a short temperature spike of up to 28°C. As of 10:18 CEST, on call duty received alarms about raising temperature within the cold aisle. Our first mitigation in such a case is the checking of cold aisle sensors and increase of supply temperature - either through other aircon units or through direct free cooling. We choose to increase the supply of cold air from direct free cooling due to current low outside temperatures. The system was now adding 20.000m³/h additional cold air to the datacenter. At the same time, we reduced temperatures of remaining aircon units. We were able to keep temperatures at a normal range of about ~25°C across the cold aisle of RR.A. On-Site personnel qualified to work on electrical circuits was called at 10:21 CEST and started investigating at 10:26 CEST the same at the outside unit, identifying a supply issue. We escalated the same further to the air conditioning company, who’s responsible for the stable operation of the same. The unit was repaired and cooling restarted. We are aware that technical defects can cause cooling abruptly to stop. For that reason, Tornado Datacenter operates multiple isolated air condition units, supplying cold air. Due to the increased electrical demands of colocation customers, we started to implement a n\+1 redundant cold water system. The same works in concert with our in-house engineered air-water heat-exchanger architecture, which is about to be taken live in the next week.