The RPS is calculated with a counter on the number of request divided by the test time elapsed.
This is wrong because if the loadtest is doing long pending stuff (as websocket tests) that are not related to requests, then the RPS is not at all accurate.
I think we should count the number of requests devided by the sum of requests time elapsed.
Then we would have a more accurate RPS that is not related to the loadtests duration.
The RPS is calculated with a counter on the number of request divided by the test time elapsed.
This is wrong because if the loadtest is doing long pending stuff (as websocket tests) that are not related to requests, then the RPS is not at all accurate.
I think we should count the number of requests devided by the sum of requests time elapsed.
Then we would have a more accurate RPS that is not related to the loadtests duration.