After struggling a few days with that Websphere-Cookie issue, I finally managed to run a few gatling test.
The requests are correctly sent, (and already allowed us to pin-point a multithread bug in the application =D ),
and the responses appear to be OK ; no exception in my server log, and Gatling extracts the expected values from the response body without issue.
However, the report seems wrong : all the reported response times are under 1ms.
For example, I have one request that loads and displays a shitload of data ; for some cases, it takes between 2 and 4 seconds to load everything. (time recorded inside Firefox with Firebug)
But when I run with Gatling the same request on these same cases, reports indicate that the response time is under 1ms, 0.04ms according to the Reponse Time during Simulation graph.
Has anyone ever run into that kind of issue?
What could have I missed or misunderstood?
Even if 1ms seems actually very small, just a small guess: Gatling works at the network level. So, if your data takes actually not much time to download, but quite a lot to render inside the browser, that could be a part of the explanation?
I’ve thought about that one too, especially since we have some issues with the slowness of the Javascript on some version of IE,
but Firebug does display 2~3s “waiting” and a only few milliseconds of javascript processing.
My server logs also displays way more than a few ms of processing,