The setup of WildFly in the jberet-tck-runner project
I have written a JBeret blog post describing the CI task setup of the jberet-tck-runner:
I’d like to add some notes here about the WildFly server setup in the CI in this blog post. Firstly the shared WildFly build is added in the rawhide CI:
wildfly-build:
uses: wildfly/wildfly/.github/workflows/shared-wildfly-build.yml@main
with:
wildfly-branch: "main"
wildfly-repo: "wildfly/wildfly"
The above setup will build the WildFly main branch, and it will produce a zip file of the WildFly maven repo. The CI task will extract the zip:
- name: Extract WildFly Maven Repo
shell: bash
run: |
tar xvzf wildfly-maven-repository.tar.gz -C ~
In the pom.xml of the project, the wildfly-maven-plugin can use the above extracted WildFly repo for provision:
<plugin>
<groupId>org.wildfly.plugins</groupId>
<artifactId>wildfly-maven-plugin</artifactId>
<version>${version.wildfly-maven-plugin}</version>
...
</plugin>
And the WildFly version is injected during the running process:
USE_PROFILE="${USE_PROFILE}" \
WFLY_VER="${WFLY_VER}" \
JBERET_VER="${jberet_ver}" \
BATCH_TCK_DIR="${BATCH_TCK_DIR}" \
./run-wildfly-ci.sh
The above WFLY_VER is fetched and injected in the CI task:
- name: Run Rawhide Tests With Default WildFly
run: WFLY_VER=$\{\{needs.wildfly-build.outputs.wildfly-version\}\} USE_BRANCH=$\{\{ matrix.use_branch \}\} ./run-tck-rawhide.sh
The variable needs.wildfly-build.outputs.wildfly-version is produced by the wildfly/wildfly/.github/workflows/shared-wildfly-build.yml.