behave-docker-parallel

---

TODO

• implement scenarios that actually run in a browser
• run python in worker container as non-root

Requirements

• docker
• docker-compose
• pre-commit
• (optional) virtualenvwrapper or pipenv

Create a virtualenv with virtualenvwrapper:

mkvirtualenv -p python3.7 bdp

Then for development purposes install all requirements with:

make requirements

Updating requirements

• Add a new dependency to requirements.in without version pinning
• Then compile new requirements.txt with pip-compile

make compile_requirements

Start Hub & Browsers

Start Selenium Hub and 1 Firefox, 1 Chrome & 2 workers (1 for each browser)

make containers

You can also call docker-compose directly:

docker-compose up --build --scale worker=2

Alternatively you can spin up multiple workers and instances of FF and/or Chrome:

docker-compose up --build --scale firefox=3 --scale chrome=3 --scale worker=6

Run tests

make tests

Management & Monitoring Consoles

Once all services are up & running you can access:

• Selenium Grid Console http://localhost:4444/grid/console
• Redis Commander (Redis monitoring tool) http://localhost:8081/

Allure reports

• Install allure
• or if you're e.g. on Fedora then install NPM wrapper around allure-commandline
  • npm install -g allure-commandline --save-dev

Once all tests are done, the you'll find allure result files in browser specific folders, e.g.: chrome_results & firefox_results. Some of those JSON result files might contain the same historyId & uuid values. If you decide to simply merge those folders and generate an allure report from them, then some of the test results might "disappear" as they will be overwritten by test with the same historyId and/or uuid. In order to address this issue please use make results (or update_results.py script). It will prepend a custom prefix (browser name) to aforementioned fields and to the test name. This turn will make test results unique.

To update result files and move them to the ./results folder use:

make results

Then you can either serve the report directly from result files with:

make serve

or you can generate a standalone html report with:

make report

Once allure starts up the web server it will open in your default browser, and thanks to using the prefix, you'll see the tests nicely

Tips

List all scenario names

make scenarios

Finding duplicated scenario names

In order to run all of your scenarios, they have to have a unique name. If that's not the case then some of the scenarios might not be executed and/or their results won't be visible in the allure report.

To find duplicated scenario names use:

make find_duplicated_scenario_names
--- /dev/fd/63	2019-10-07 15:43:06.897083292 +0100
+++ /dev/fd/62	2019-10-07 15:43:06.898083296 +0100
@@ -7,4 +7,3 @@
 Updating group
 Updating user
 run a simple test
-run a simple test
make: *** [makefile:25: find_duplicated_scenario_names] Error 1

Checking health status of Selenium Hub & Browsers

You can check the health status of selenium-hub or selected browser with:

docker inspect --format='{{json .State.Health.Status}}' selenium-hub
# or
docker inspect --format='{{json .State.Health.Status}}' #BROWSER_INSTANCE_ID

or you can use a "get'em all" one-liner that displays health statuses for all browser instances and selenium-hub in one go:

docker inspect --format '{{ .Name }} -> {{ .State.Health.Status }}' \
    $(docker ps -a --format "table {{.Names}}\t{{.ID}}" |\
    grep "chrome\|firefox\|hub" |\
    awk '{printf $NF" "}')

btw. you can always create a handy alias for that one-liner:

alias hh='docker inspect --format "{{ .Name }} -> {{ .State.Health.Status }}" $(docker ps -a --format "table {{.Names}}\t{{.ID}}" | grep "chrome\|firefox\|hub" | awk '"'"'{printf $NF" "}'"'"')'