Detect hanging handlers to enable mitigating actions from this state

[ramonsmits]

Describe the feature.

Message processing could slow down or even halt when handlers never complete (hang). For example, handlers that deadlock or infinitely wait on a state change that will never happen and the code is not applying any timeout logic or (cooperative) cancellation.

Ideally, the framework should detect such hanging or extremely long-running handlers. After detecting such handlers the endpoint should at minimum log about this. Likely the endpoint will be considered to be in an unhealthy state. It could raise a critical error so that its host can decide to terminate and restart to resolve such persistent issues.

Related

The following community project provides partial solutions to this problem

• Reports long-running handlers: https://github.com/ramonsmits/NServiceBus.SimpleMonitoring
• Raises a critical error if there is no activity for a configurable duration: https://github.com/ramonsmits/NServiceBus.GuardNoActivity

Additional Context

No response