b0610cb63c
|
3 years ago | |
|---|---|---|
| .. | ||
| soak | ||
| README.md | ||
| ansible_install.py | ||
| ansible_tests.py | ||
| azure-pipelines-steps.yml | 3 years ago | |
| azure-pipelines.yml | 3 years ago | |
| ci_lib.py | 3 years ago | |
| debops_common_install.py | ||
| debops_common_tests.py | 3 years ago | |
| localhost_ansible_install.py | ||
| localhost_ansible_tests.py | 3 years ago | |
| mitogen_install.py | ||
| mitogen_py24_install.py | ||
| mitogen_py24_tests.py | ||
| mitogen_tests.py | ||
README.md
.ci
This directory contains scripts for Continuous Integration platforms. Currently Azure Pipelines, but they will also happily run on any Debian-like machine.
The scripts are usually split into _install and _test steps. The _install
step will damage your machine, the _test step will just run the tests the way
CI runs them.
There is a common library, ci_lib.py, which just centralized a bunch of
random macros and also environment parsing.
Some of the scripts allow you to pass extra flags through to the component
under test, e.g. ../../.ci/ansible_tests.py -vvv will run with verbose.
Hack these scripts until your heart is content. There is no pride to be found here, just necessity.
ci_lib.run_batches()
There are some weird looking functions to extract more paralellism from the
build. The above function takes lists of strings, arranging for the strings in
each list to run in order, but for the lists to run in parallel. That's great
for doing setup.py install while pulling a Docker container, for example.
Environment Variables
TARGET_COUNT: number of targets fordebops_run. Defaults to 2.DISTRO: themitogen_tests need a target Docker container distro. This name comes from the Docker Hubmitogenuser, i.e.mitogen/$DISTRO-testDISTROS: theansible_tests can run against multiple targets simultaneously, which speeds things up. This is a space-separated list of DISTRO names, but additionally, supports:debian-py3: when generating Ansible inventory file, setansible_python_interpretertopython3, i.e. run a test where the target interpreter is Python 3.debian*16: generate 16 Docker containers running Debian. Also works with -py3.