Hi Peter,
Off the top of my head, there are two approaches which may meet your requirement…
One approach is to use an incident handler. When the retry policy is exceeded, the engine will raise an incident and suspend the execution. You could consider a custom incident handler to perform a notification activity (see here)
An alternate approach could be to take remedial action in the process itself. Have a look at this thread - fail fast. If you look at the sample code, you can see how you can tell you are in a job executor context and the number of retries remaining…
regards
Rob