Have you ever felt annoyed because of the long waiting time for receiving test results? Maybe after a few hours, you’ve figured out that there had been a network connection issue in the middle of testing, and half of the results can go to the trash? That may happen when your tests are dependent on each other or when you have plenty of them and execution lasts forever. It’s quite a common issue. But there’s actually a solution that can not only save your time but also your money – parallelization in the Cloud.
How It Started
Developing UI tests for a few months, starting from scratch, and maintaining existing tests, I found out that it has become something huge that will be difficult to take care of very soon. An increasing number of test scenarios made every day led to bottlenecks. One day when I got to the office, it turned out that the nightly tests were not over yet. Since then, I have tried to find a way to avoid such situations.
A breakthrough was the presentation of Tomasz Konieczny during the Testwarez conference in 2019. He proved that it’s possible to run Selenium tests in parallel using AWS Lambda. There’s actually one blog that helped me with basic Selenium and Headless Chrome configuration on AWS. The Headless Chrome is a light-weighted browser that has no user interface. I went a step forward and created a solution that allows designing tests in the Behavior-Driven Development process and using the Page Object Model pattern approach, run them in parallel, and finally – build a summary report.
Setting Up the Project
The first thing we need to do is signing up for Amazon Web Services. Once we have an account and set proper values in credentials and config files (.aws directory), we can create a new project in PyCharm, Visual Studio Code, or in any other IDE supporting Python. We’ll need at least four directories here. We called them ‘lambda’, ‘selenium_layer’, ‘test_list’, ‘tests’ and there’s also one additional – ‘driver’, where we keep a chromedriver file, which is used when running tests locally in a sequential way.
In the beginning, we’re going to install the required libraries. Those versions work fine on AWS, but you can check newer if you want.
requirements.txt
allure_behave==2.8.6
behave==1.2.6
boto3==1.10.23
botocore==1.13.23
selenium==2.37.0
What’s important, we should install them in the proper directory – ‘site-packages’.
We’ll need also some additional packages:
All those things will be deployed to AWS using Serverless Framework, which you need to install following the docs. The Serverless Framework was designed to provision the AWS Lambda Functions, Events, and infrastructure Resources safely and quickly. It translates all syntax in serverless.yml to a single AWS CloudFormation template which is used for deployments.
Architecture – Lambda Layers
Now we can create a serverless.yml file in the ‘selenium-layer’ directory and define Lambda Layers we want to create. Make sure that your .zip files have the same names as in this file. Here we can also set the AWS region in which we want to create our Lambda functions and layers.
serverless.yml
service: lambda-selenium-layer
provider:
name: aws
runtime: python3.6
region: eu-central-1
timeout: 30
layers:
selenium:
path: selenium
CompatibleRuntimes: [
"python3.6"
]
chromedriver:
package:
artifact: chromedriver_241.zip
chrome:
package:
artifact: headless-chromium_52.zip
allure:
package:
artifact: allure-commandline_210.zip
resources:
Outputs:
SeleniumLayerExport:
Value:
Ref: SeleniumLambdaLayer
Export:
Name: SeleniumLambdaLayer
ChromedriverLayerExport:
Value:
Ref: ChromedriverLambdaLayer
Export:
Name: ChromedriverLambdaLayer
ChromeLayerExport:
Value:
Ref: ChromeLambdaLayer
Export:
Name: ChromeLambdaLayer
AllureLayerExport:
Value:
Ref: AllureLambdaLayer
Export:
Name: AllureLambdaLayer
Within this file, we’re going to deploy a service consisting of four layers. Each of them plays an important role in the whole testing process.
#aws #aws lambda #qa and software testing #serveless
