Spend some time designing a highly debuggable and consistent system around fault point management. The idea is that there is a consistent behaviour around each fail point, which is predictable and observable (logs). Think about a logging mechanism when a fail point:

• is hit
• is enabled/disabled
• is configured to execute at random frequency
• is set to a value

Consider what might be required as part of hitting a fault point, eg: do we generate a core differently if we were to panic, do we need to provide the fault point writer means to output a writer-defined log, etc. Also, make sure we do not overwhelm the logging system.

The ticket is limited to doing the design work alone.