Open kashish-230997 opened 4 years ago
Could you explain your use case in a little more detail? RPS reported by Artillery is the number of requests served per second, not requests made per second. As such, the number will depend on many factors outside of Artillery's control, such as network conditions and most importantly the speed of the system under test.
There are tools that can send a constant number of requests per second (e.g. Vegeta) but that model can only work when your scenarios are composed of completely independent requests, such as sending the same GET
request to a webserver. When you need to be able to build scenarios composed of multiple dependent requests (e.g. login first, and then use the auth token to make another request), that model does not work. Artillery is designed for testing applications and systems with complex multi-step flows, not the simple use case of testing a web server. If your use-case is closer to the first one described, you should consider using Vegeta instead.
Actually we were using Locust tool for performance testing but we were facing issue of RPS value as it was not coming within the range so we are trying to use Artillery tool to have proper RPS(request per second) to get better performance. but we are not able to provide range for RPS or control it according to requirement.
My query is somewhat similar. I am using Artillery for load testing our API, but somehow - the RPS starts to decline around 140-150. Thereafter, we start to see bunch of other errors as well.
Is there any way to put increased number of RPS. The target here is 5k, and we are not able to go beyond 150!
we need to calculated and control RPS dynamically in artillery to improve our load testing.