Load test runner using Nightwatch scripts

[reallymello]

Description

I would love to leverage some of our existing functional Nightwatch testing scripts at load and have the results compiled into a load test report. This is something Artillery.io is doing with modified Playwright scripts as a different way of generating load than creating batches of API requests.

Using a UI script in the performance test would put a more accurate load on the system under test and would allow us to report on actual user experience navigating the system under test. It would save a tremendous time over decomposing the page requests into JMeter or Artillery requests which could also deviate over time with reality as the site changes.

Suggested solution

I feel that Nightwatch has the architecture in order to support this already in that it can run test scripts in parallel. What I propose we add is the ability to

• Specify which scripts to run
• How long to continuously run the scripts
• How many users / threads should run the scripts over that time period
• Optionally, ability to have phases to ramp up and down user load and run different scripts at different phases
• Output a performance test report averaging the normal step timings with perhaps the browser metrics from the .getPerformanceMetrics method

Alternatives / Workarounds

Copy and paste existing tests multiple times and run them in parallel with a custom reporter that would provide me average times over the runs of the same tests

Additional Information

No response

[github-actions]

Thank you for setting this as an enhancement. One of the product folk will triage this again to help see when we can fit this in an upcoming sprint.