This change eliminates the need to install the connector on each controlled node, as `pymysql` version 1.1.1 is now included. As a result, we can safely assume its availability, thus simplifying the testing process. Also, I managed to remove the need for pre-built test containers. We now use the default test containers from ansible-test.
5.5 KiB
Tests
This collection uses GitHub Actions to run ansible-test to validate its content. Three type of tests are used: Sanity, Integration and Units.
The tests covers plugins and roles (no role available yet, but tests are ready) and can be found here:
- Plugins: .github/workflows/ansible-test-plugins.yml
- Roles: .github/workflows/ansible-test-roles.yml (unused yet)
Everytime you push on your fork or you create a pull request, both workflows runs. You can see the output on the "Actions" tab.
Integration tests
You can use GitHub to run ansible-test either on the community repo or your fork. But sometimes you want to quickly test a single version or a single target. To do that, you can use the Makefile present at the root of this repository.
For now, the makefile only supports Podman.
Requirements
- python >= 3.8 and <= 3.10
- make
- podman
- Minimum 15GB of free space on the device storing containers images and volumes. You can use this command to check:
podman system info --format='{{.Store.GraphRoot}}'|xargs findmnt --noheadings --nofsroot --output SOURCE --target|xargs df -h --output=size,used,avail,pcent,target
- Minimum 2GB of RAM
Makefile options
The Makefile accept the following options
-
local_python_version
- Mandatory: false
- Choices:
- "3.8"
- "3.9"
- "3.10"
- Description: If
Python -V
shows an unsupported version, use this option and choose one of the version available on your system. Usels /usr/bin/python3*|grep -v config
to list them.
-
ansible
- Mandatory: true
- Choices:
- "stable-2.15"
- "stable-2.16"
- "stable-2.17"
- "devel"
- Description: Version of ansible to install in a venv to run ansible-test
-
db_engine_name
- Mandatory: true
- Choices:
- "mysql"
- "mariadb"
- Description: The name of the database engine to use for the service containers that will host a primary database and two replicas.
-
db_engine_version
- Mandatory: true
- Choices:
- "5.7.40" <- mysql
- "8.0.31" <- mysql
- "10.4.24" <- mariadb
- "10.5.18" <- mariadb
- "10.6.11" <- mariadb
- Description: The tag of the container to use for the service containers that will host a primary database and two replicas. Do not use short version, like
mysql:8
(don't do that) because our tests expect a full version to filter tests precisely. For instance:when: db_version is version ('8.0.22', '>')
. You can use any tag available on hub.docker.com/_/mysql and hub.docker.com/_/mariadb but GitHub Action will only use the versions listed above.
-
python
- Mandatory: true
- Choices:
- "3.8"
- "3.9"
- "3.10"
- Description: The python version to use in the controller (ansible-test container).
-
target
- Mandatory: false
- Choices:
- "test_mysql_db"
- "test_mysql_info"
- "test_mysql_query"
- "test_mysql_replication"
- "test_mysql_role"
- "test_mysql_user"
- "test_mysql_variables"
- Description: If omitted, all test targets will run. But you can limit the tests to a single target to speed up your tests.
-
keep_containers_alive
- Mandatory: false
- Description: This option keeps all tree databases containers and the ansible-test container alive at the end of tests or in case of failure. This is useful to enter one of the containers with
podman exec -it <container-name> bash
for debugging. Rerunning the tests will overwrite the 3 databases containers so no need to kill them in advance. But nothing will kill the ansible-test container. You must do that usingpodman stop
andpodman rm
. Add any value to activate this option:keep_containers_alive=1
-
continue_on_errors
- Mandatory: false
- Description: Tells ansible-test to retry on errors and also continue on errors. This is the way the GitHub Action's workflow runs the tests. This can be used to catch all errors in a single run, but you'll need to scroll up to find them. Add any value to activate this option:
continue_on_errors=1
Makefile usage examples:
# Run all targets
make ansible="stable-2.12" db_engine_name="mysql" db_engine_version="5.7.40" python="3.8"
# A single target
make ansible="stable-2.14" db_engine_name="mysql" db_engine_version="5.7.40" python="3.8" target="test_mysql_info"
# Keep databases and ansible tests containers alives
# A single target and continue on errors
make ansible="stable-2.14" db_engine_name="mysql" db_engine_version="8.0.31" python="3.9" target="test_mysql_query" keep_containers_alive=1 continue_on_errors=1
# If your system has an usupported version of Python:
make local_python_version="3.8" ansible="stable-2.14" db_engine_name="mariadb" db_engine_version="10.6.11" python="3.9"
Run all tests
GitHub Action offer a test matrix that run every combination of Python, MySQL, MariaDB and Connector against each other. To reproduce this, this repo provides a script called run_all_tests.py.
Examples:
python run_all_tests.py
Add a new Python or Database version
You can look into .github/workflows/ansible-test-plugins.yml
- Add your version in the matrix of .github/workflows/ansible-test-plugins.yml. You can use run_all_tests.py to help you see what the matrix will be. Simply comment out the line
os.system(make_cmd)
before runing the script. You can also addprint(len(matrix))
to display how many tests there will be on GitHub Action.