Is it possible to exclude failed requests from the max. time calculation?

Under load, some of our services are taking an inordinate amount of time to respond, or simply never responding. Thus we end up with max. response times over 60 seconds. Since these requests have failed, is there a way to exclude their response times from the max response time calculation?


That is currently not possible.

What you really want to know is the 99th percentile anyway, since that is what your users will be noticing.