The issue is related to the improper handling of exceptional conditions and uncontrolled resource consumption in Apache Tomcat when processing an HTTP/2 stream. This led to a miscounting of active HTTP/2 streams, which in turn led to the use of an incorrect infinite timeout, allowing connections to remain open that should have been closed. It is estimated that over 50,300 services are potentially affected.