We experienced an issue with in-memory storage where temporary files were cleaned up slower than usual at the end of a request. This caused a buildup of unused data, leading to memory issues and reduced performance on October 1st from 9:30 AM to 11:00 AM CEST. During this period, requests to the DocHorizon model capturing services may have taken longer than usual or resulted in timeout errors.
The issue has been fully fixed, and all services are functioning normally again.