Restore pipefail in Tests workflow to fail on non-zero exit codes

The hashicorp/setup-terraform action with terraform_wrapper enabled
(default) overrides the bash shell from 'bash -eo pipefail' to
'bash -e', stripping pipefail. This caused pipe commands like
'ansible-playbook ... | tee ...' to swallow non-zero exit codes,
letting the CI job pass despite Ansible test failures.

Fix by disabling terraform_wrapper (not needed since we don't use
terraform outputs in GitHub Actions) and adding explicit set -o
pipefail to steps that use pipes as defense-in-depth.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: szachovy

.github/workflows/tests.yml

@@ -32,6 +32,7 @@ jobs:
       - uses: "hashicorp/setup-terraform@v3"
         with:
           terraform_version: "1.0.10"
+          terraform_wrapper: false
       - uses: "actions/setup-python@v3"
         with:
           python-version: "3.10.12"
@@ -49,6 +50,7 @@ jobs:
 
       - name: "Build and pre-load service images into test containers"
         run: |
+          set -o pipefail
           build_and_load() {
             local image_name="$1"
             local build_context="$2"
@@ -71,6 +73,7 @@ jobs:
 
       - name: "Run Ansible testsuite"
         run: |
+          set -o pipefail
           ansible-playbook --inventory inventory.yml deploy.yml \
             2>&1 | tee -a /tmp/superset_cluster_testsuite.log
         working-directory: "${{ env.ANSIBLE_WORKING_DIRECTORY }}"

CHANGELOG.md

@@ -44,6 +44,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 * Made `create_directory()` idempotent to support deployment re-runs. (#35)
 * Added default inventory directive to `ansible.cfg`. (#42)
 * Fixed Publish workflow `write_package` permission denied by adding OCI source labels for GHCR repository linking. (#99)
+* Restore `pipefail` in Tests workflow to propagate non-zero exit codes through pipes.
 
 ## 1.0 - 2024-10-13