AWS Builder Center

What unit tests missed: testing a serverless AWS app on the real cloud

I'm a first-year student, and on our hackathon project, Deadman, I owned testing. Deadman applies an AWS security-group change and undoes it on a timer unless you confirm. Our unit tests all passed, and the real cloud still found problems they never could.

Student
What real end-to-end testing found: an API authorizer that returned a 500 for every request; a website still showing an old build because CloudFront had cached it; a lock left behind by a failed request, which blocked the next attempt; and the big one, an auto-revert that never fired because the schedule pointed at a function name that did not exist.
How I found each one: I drove the live website and API myself, with a demo server next to the browser, so I could see the server go dark and, once we fixed the bug, come back on its own. I recorded timestamps, and my teammates read the CloudWatch logs to find causes.
What I would tell another beginner: write down what each check actually proves and what it does not. Our verification log marks each check as passed, failed, or not run, so nobody fools themselves. And test the boring failure cases on purpose: a wrong token, a busy resource, a timer that fires late.
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article