I a have a design question about the way Gatling is sending the next request.
I have a scenario that is using the default Gatling checking, that is based on the value of the HTTP status code.
But my application under test always returns 200 in case of application errors so I have to add an extra check based on the xpath extractor as my response is XML.
But as I’m very focused on throughput, as It will add non negligeable processus on the Gatling side, I’m wondering is the next request is send after this processing has been done or immediately after the current request has been received and before the check is performed.
Here’s what’s done sequentially:
- send request
- receive all response chunks
- perform checks
- save time spent into performing checks into the session (this time will be substracted from the next pauses)
- perform next action (another request, a pause, whatever’s next in your scenario)
We can’t perform the next action before performing checks because of error handling structures such as tryMax.
If you use XPath, have you checked the VTD-XML based extension? It should be faster and more memory efficient.
Hope that helps.
2012/10/28 Jeff MAURY <firstname.lastname@example.org>
I’m gonna try this extension and see the impact on the throughput.
Thanks for the pointer