### Description of the change
This adds an environment for unit testing our bash scripts, using [BATS](https://github.com/bats-core/bats-core).
It implements first tests for `config_environment.sh`.
### Benefits
Writing unit tests for bash scripts documents the expected behavior and allows it being a quality gate in our CI.
### Possible drawbacks
Not everyone is familiar with this approach and unit testing framework. Me neither, it took me some hours to get into it.
### Applicable issues
- Related to #691 where an issue in `config_environment.sh` was detected. It doesn't fixes it yet. This will be a dedicated Pull Request.
### Additional information
I've verified that the changes for Renovate are indeed working.
You may wonder why there is only one `run $PROJECT_ROOT/scripts/init-containers/config/config_environment.sh` and many `run execute_test_script` calls.
Usually, testing a script itself would be executing `run $PROJECT_ROOT/scripts/init-containers/config/config_environment.sh`. You then can assert the exit code and other things. Since the `config_environment.sh` exports environment variables and we are not able to access them from outside a `run` execution, the function `execute_test_script` wraps our script execution between environment comparison. Doing so allows us capture environment variables that were added/removed during script execution.
Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/724
Reviewed-by: pat-s <pat-s@noreply.gitea.com>
Co-authored-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Co-committed-by: justusbunsi <sk.bunsenbrenner@gmail.com>
### Description of the change
This is a preparation for #724 (bash script testing) which would be too complex to review if done there. I haven't changed the script contents.
### Benefits
Better IDE support, possibility for shell checks, tests.
### Possible drawbacks
-
### Additional information
I only extracted files that are native bash scripts. There are others, being a mix of Helm templating and shell scripts. Those scripts must be cleared first before extracting them as well.
Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/751
Co-authored-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Co-committed-by: justusbunsi <sk.bunsenbrenner@gmail.com>
### Description of the change
This addresses 2 bugs regarding configured runner labels:
- `Cannot find: node in PATH` due to a default label that matches the runner host, not the docker-in-docker environment
- After changing the runner config, users have to manually replace the StatefulSet Pod of the runner
### Benefits
- Out-of-the-box functioning node environment
- Auto-updated runner config on changes in values
### Applicable issues
- fixes https://gitea.com/gitea/helm-chart/issues/728
- fixes https://gitea.com/gitea/helm-chart/issues/725
### Additional information
With removing the default labels from `values.yaml`, the act runner defaults will be used for new runners which fixes the "missing node" issue.
I had issues replacing the labels once the Chart defaults were used, so users might have to replace their runner manually if this is an issue.
### Checklist
- [x] Templating unittests are added
Signed-off-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/739
Co-authored-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Co-committed-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Right now, the generated `.prov` file is not uploaded along with the actual `.tgz` file. This makes it impossible to verify our Helm Charts.
In addition, we only sign the old-fashioned `.tgz` file, not the OCI-based releases on DockerHub.
The incentive to do this very commit is an expired GPG key that prevents our release.
Signed-off-by: justusbunsi <sk.bunsenbrenner@gmail.com>
The current README content is a result of running markdownlint with simple fixes enabled.
It shouldn't fix _any_ issues, even after generating the parameter section.
Unfortunately, this is not the case right now.
This resolves the long-running dispute of both tools. 😉
---
Since I already modified the `values.yaml`, I decided to include a change regarding code signing.
Otherwise, it may confuse users after merging #596.
Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/627
Reviewed-by: pat-s <pat-s@noreply.gitea.com>
Co-authored-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Co-committed-by: justusbunsi <sk.bunsenbrenner@gmail.com>
### Description of the change
With https://github.com/go-gitea/gitea/pull/28390, Gitea 1.21.2 introduced warning log output within the result of `gitea admin <subcommand>` and therefore affects the current provisioning script.
That script previously assumed a clean result set and was therefore doomed to fail at _some_ point.
This introduces output sanitizing to trim such logs above the actual result table.
### Applicable issues
- fixes#589
### Additional information
The non-sanitized output were only an issue for admin account provisioning, and only when the username matched one of these words (in case of #589 it was `gitea`):
```text
.../setting/security.go:168:loadSecurityFrom() [W] Enabling Query API Auth tokens is not recommended. DISABLE_QUERY_AUTH_TOKEN will default to true in gitea 1.23 and will be removed in gitea 1.24.
```
LDAP and OAuth sources were not affected by this particular log line, but also processed non-sanitized result sets. Changing their code is a precaution.
Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/590
Reviewed-by: pat-s <pat-s@noreply.gitea.com>
Co-authored-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Co-committed-by: justusbunsi <sk.bunsenbrenner@gmail.com>
### Description of the change
This adds a `.vscode` folder with recommended extensions and some useful settings like unittest schema validation.
The `.vscode` folder is already helm ignored during packaging.
### Possible drawbacks
We would have to be careful about PR changes in that directory.
Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/536
Co-authored-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Co-committed-by: justusbunsi <sk.bunsenbrenner@gmail.com>
### Description of the change
We are affected by a regression of a Helm bug from May 2023. I've tested
the Helm versions 3.13.1, 3.13.0 and 3.12.3. Both 3.13.x are affected.
3.12.3 works. So let's downgrade and drop the docker login in PR builds.
I've also switched the `apt install helm` with an official `alpine/helm`
image I am using at work. Pinning the helm version and receiving updates
helps us identifying such issues in the future.
For the release workflow I was a bit more reluctant with changes, since
I cannot easily test them. That's why I just pinned the Helm version.
Renovate will provide one PR changing both files because it's the same
dependency (alpine/helm) from the same datasource (docker).
https://github.com/helm/helm/issues/12062
### Applicable issues
- implicitly fixes#527
Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/535
Reviewed-by: pat-s <pat-s@noreply.gitea.com>
Co-authored-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Co-committed-by: justusbunsi <sk.bunsenbrenner@gmail.com>
### Description of the change
This adds a new values object `serviceAccount`, that allows creating a dedicated ServiceAccount with the Helm Release into the cluster. It supports all common options like labels, annotations, name override (or referring to an externally created ServiceAccount), auto-mount token, image pull secrets.
It supersedes the stale PR #357.
### Benefits
Users can deploy Gitea with more fine-tuned security settings.
### Applicable issues
- related to #448
### Additional information
I've bumped the helm-unittest plugin in the CI build, to be able to use the `exists` and `notExists` feature in the new tests.
### Checklist
- [x] Parameters are documented in the `values.yaml` and added to the `README.md` using [readme-generator-for-helm](https://github.com/bitnami-labs/readme-generator-for-helm)
Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/451
Reviewed-by: pat-s <pat-s@noreply.gitea.com>
Co-authored-by: justusbunsi <sk.bunsenbrenner@gmail.com>
Co-committed-by: justusbunsi <sk.bunsenbrenner@gmail.com>
