In support cases, we often have to diagnose cause of hung or excessively slow requests, but no helpful data is captured on this by default. Customers must manually capture thread dumps but often don't know to do so. This delays RCAs and resolutions as we have to teach customers what to enable/capture and wait for a recurrence.

An improvement would be if our default server.xml already starts with StuckThreadDetectionValve enabled. This could help us gain some data and possibly enough to close a case from the logging already generated.