Wrap Up Selenium: The Feedzai Approach to Run Thousands of Tests in a Distributed Way
At Feedzai, strong quality requirements are part of the overall creation and release protocol for our products is enforced by us on behalf of our clients. Thus, we spend a considerable portion of our efforts in the test automation phase using Selenium, so that we can confidently say that the top layer of our product is solid, even on a nightly build schedule.
Feedzai, analyzes more than $2 billion in purchases in real-time each day, all over the world.
In order to run roughly 1000 selenium tests once every 2/3 commits, we have developed a test framework that creates deployments of our product using docker. This framework starts and configures all the applications/containers required to have working data centers with out product. The web driver is then connected to a docker container running a browser, and the testing is executed.
The isolation provided by a docker allows us to run more than 20 tests in parallel in CI. This means that a single server withstands tremendous system loads, which allows us to focus on and stress known test instability points associated with selenium. Thus, we have developed a selenium wrapper that encapsulates the required polling and retries.
Finally, we’ve organized our selenium codebase in a per screen basis. This means that for every screen in our web application, there’s an easy (single line) way of performing an operation using the UI. This leads to awesome selenium tests that are both easy to read and easy to write.
In this Test our Coffee session we will be presenting our approach and discussing it with other participants, who share common problems, and understand what steps each one of us takes in order to “Wrap Up Selenium”.
Bernardo is a young and passionate Software Engineer in Test who joined Feedzai in early 2016.
He was responsible for migrating the selenium test base from a selenium-grid approach to the current docker based approach, which allows us to run the builds in more general purpose servers, thus increasing the degree of parallelism both within and among CI builds.
Ricardo is connected to Quality Assurance and Software Testing since the beginning of his IT career and have been involved in multiple software projects with high complexity.
He tries to cross join the technical knowledge in multiple testing tools with his natural “bug-hunter” skills in a perfect synergy.
Since he joined Feedzai, he was responsible to validate and automate tests for a whole new Data Science platform, that helps Data Scientists to create the best models for fraud detection. While automating each test scenario, even the most complex ones, tries to keeps focused and draw the best architecture in order for the tests to follow the principle: “Keep it simple, stupid”.
DATE AND TIME:
31 January, 2017 – 18:30h to 20:00h
Rua Pedro Nunes
IPN – Instituto Pedro Nunes
This event will be hosted by Feedzai and PTSQB – Portuguese Association of Software Testing, an official member from ISTQB (International Software Testing Qualifications Board).
In order to enroll check the details below (in Portuguese):
INSCRIÇÕES LIMITADAS! INSCREVA-SE JÁ!
Inscrições: [email protected]
Associados da PSTQB: Grátis
Não associados da PSTQB: 10.00€
(O pagamento tem de ser efetuado por transferência antes do dia do evento, para o IBAN PT50 0035 0001 00033933330 69 indicando o primeiro e último nomes)
Ainda não é associado? Torne-se já!
Saiba mais em www.pstqb.pt