By running our end-to-end (playwright) testing suite in a serverless/lambda environment, we have managed to make e2e testing less of a deployment bottleneck. This approach has also made it easier for engineers to discover and combat flakiness up-front.
Silver bullet? Not so much. Flakey tests are gonna flake, and it is still difficult to run tests on components in isolation. Not to mention that you still need to maintain a stable environment to run them against and provide a way for engineers to test their changes in that environment.
With that said, they’re still a fantastic tool when used in the right way! They help us catch some of the most unexpected failures before they reach production.
In this talk we will consider what an end-to-end testing strategy looks like in a world where your e2e tests are no longer expensive in terms of hardware resources or time. Should this change the way we rate e2e tests? Does this turn the testing pyramid upside down?
Key takeaways:- What problems are and are not solved by running e2e tests in serverless/lambda environments
- How serverless environments shift the cost of testing
- How this approach can help developers discover and combat flakiness upfront