26 lines
1.5 KiB
Markdown
26 lines
1.5 KiB
Markdown
# GitHub Actions
|
|
## Background
|
|
We want to make attempts to run our integration tests in the same manner wether using GitHub actions or on a developers's machine locally. For this reason, the tests are curated to run using container images. As of this writing, two images exist which we would like to test against:
|
|
- quay.io/ansible-product-demos/apd-ee-24:latest
|
|
- quay.io/ansible-product-demos/apd-ee-25:latest
|
|
|
|
These images are built given the structure defined in their respective EE [definitions][../execution_environments]. Because they differ (mainly due to their python versions), each gets some special handling.
|
|
|
|
## Troubleshooting GitHub Actions
|
|
|
|
### Interactive
|
|
It is likely the most straight-forward approach to interactively debug issues. The following podman command can be run from the project root directory to replicate the GitHub action:
|
|
```
|
|
podman run \
|
|
--user root \
|
|
-v $(pwd):/runner:Z \
|
|
-it \
|
|
<image> \
|
|
/bin/bash
|
|
```
|
|
`<image>` is one of `quay.io/ansible-product-demos/apd-ee-25:latest`, `quay.io/ansible-product-demos/apd-ee-24:latest`
|
|
It is not exact because GitHub seems to run closer to a sidecar container paradigm, and uses docker instead of podman, but hopefully it's close enough.
|
|
|
|
For the 24 EE, the python interpreriter verions is set for our pre-commit script like so: `USE_PYTHON=python3.9 ./.github/workflows/run-pc.sh`
|
|
The 25 EE is similary run but without the need for this variable: `./.github/workflows/run-pc.sh`
|