Stress / load testing

Desired behavior:

I’d like a --stress-test parameter which lets me re-run tests for a set number of hours, with a final report to indicate which tests are flaky and which are sturdy.

As an alternative, it’d be fine if your dashboard service would aggregate fails over time, allowing me to see which tests need more attention.

This feature is going to grow in importance to me personally as my test suite grows in size.

• Operating System: Linux Mint 18.3
• Cypress Version: Cypress 2.1.0
• Browser Version: Electron

Edit by @jennifer-shehane: Moved Dashboard feedback to https://portal.productboard.com/cypress-io/1-cypress-dashboard

[sneko]

I would be also interested to know if it would make sense to use Cypress tests to “load test” “all the infrastructure”?

Meaning:

• it loads the frontend
• interacts with the frontend as a real user (it’s local, but never mind it respects timing of animations and so on…)
• requests go through API
• requests go through backend + DB

Indeed being able to parallelize multiple users would help making this concrete.

Did any of you try that?

[davidjeddy]

I would like to also suggest parallel execution during --load-test. Typically I use Gatling for load testing; but Cypress and Gatling do basically the same thing: load the UI, interact, confirm.

Being able to leverage Cypress to execute a series of user actions with (for example) 10 simulated users, would cover the need for Gatling.

Stretch goal: record and display TTFB, average page load time, load bytes, HTTP code responses and counts (200,404,500,etc).