On the evening of April 27, 2026, some customers experienced intermittent issues when logging in or accessing resources. The disruption lasted approximately 5.5 hours, from 22:00 to 03:12 (GMT+2).
The issue was caused by an internal service becoming unstable, which triggered a surge of automated retry attempts that overwhelmed a supporting system. Our teams responded immediately and put the following protective measures in place:
- Added automatic safeguards to stop failing requests from piling up and spreading to other services
- Reduced wait times so requests no longer blocked the system for extended periods
- Limited the number of automatic retries to prevent traffic from multiplying during the disruption
- Stopped retries on requests that could never succeed, reducing unnecessary load
- Added safety checks to prevent one failure from triggering thousands of others
- Doubled the capacity of the affected system to handle higher demand
- Improved logging to help diagnose issues faster in the future
- Added temporary data storage so the system can continue serving customers even during disruptions
Full service was restored at 03:12 (GMT+2) on April 28, with no further issues reported. Most customer data continued to be served during the disruption, limiting the overall impact.