Currently sdks report activity_succeed_endtoend_latency
which is pretty useful to count number of activity success after all retries. There is no equivalent
of the opposite so for example activity_fail_endtoend_latency which is probably just as useful to have.

Workarounds currently is to push custom metric from workflow code (or workflow interceptor).

Ask is to add corresponding metric users can monitor for activity failures after all retries.

Just to add as note, many users think that activity_execution_failed is a metric to monitor here, but even with application failure Category that was added and also is very useful, since user does not really know when last attempt of activity is (as its regulated by server), its not really a useful alternative for this situation. Only way is to measure when server says activity failed.