菜单
KubeBlocks E2E testing infrastructure: kbcli-test.yml reusable workflow, kbcli-test-k8s.yml, cloud-e2e-*.yml, terraform-init, terraform-destroy, test-type, random-suffix. Use when adjusting kbcli-test parameters, adding E2E entry points, or changing cloud-e2e / terraform cluster lifecycle.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | e2e-testing |
| description | KubeBlocks E2E testing infrastructure: kbcli-test.yml reusable workflow, kbcli-test-k8s.yml, cloud-e2e-*.yml, terraform-init, terraform-destroy, test-type, random-suffix. Use when adjusting kbcli-test parameters, adding E2E entry points, or changing cloud-e2e / terraform cluster lifecycle. |
This skill covers generic E2E testing infrastructure: the kbcli-test reusable workflow, cloud-e2e trigger chains, and terraform cluster create/destroy.
| Workflow | Purpose |
|---|---|
.github/workflows/kbcli-test.yml | Core reusable workflow: run kbcli tests on an existing K8s cluster |
.github/workflows/kbcli-test-k8s.yml | kbcli test entry point that creates the K8s cluster first |
.github/workflows/kbcli-pre-test-k8s.yml | K8s pre-test |
.github/workflows/kbcli-test-pre.yml | kbcli pre-test preparation |
.github/workflows/cloud-e2e-engine.yml | Engine E2E on cloud |
.github/workflows/cloud-e2e-api.yml | API E2E on cloud |
.github/workflows/cloud-e2e-web.yml | Web E2E on cloud |
.github/workflows/cloud-e2e-installer.yml | Installer E2E on cloud |
.github/workflows/cloud-e2e-schedule-dev.yml | Scheduled E2E for dev environment |
.github/workflows/cloud-e2e-schedule-idc.yml | Scheduled E2E for IDC |
.github/workflows/cloud-e2e-schedule-web.yml | Scheduled Web E2E |
.github/workflows/terraform-init.yml | Create cloud K8s cluster (also in cloud-deploy) |
.github/workflows/terraform-destroy.yml | Destroy cloud K8s cluster (also in cloud-deploy) |
| Script | Purpose |
|---|---|
.github/utils/cloud_e2e_installer.sh | Install cloud E2E environment |
test-type: test type number, e.g. 0 (install), 2 (postgresql), 5 (redis).test-type-name: display name for job/reports.test-args: arguments passed to kbcli-test, commonly --service-version, --topology, --replicas.release-version: KubeBlocks version.k8s-cluster-name / artifact-name / random-suffix: cluster and artifact correlation.kbcli-test-k8s.yml
├── terraform-init.yml
└── kbcli-test.yml
kubeblocks-e2e-test.yml
├── terraform-init.yml
├── test-kubeblocks (test-type 0)
├── kbcli-test.yml (per engine/version)
├── send-message-* jobs
└── terraform-destroy.yml
kbcli-test-k8s.yml if yes, otherwise reuse kbcli-test.yml.test-type number (add in kbcli-test code if necessary).kbcli-test.yml with the correct test-args.cloud-e2e-schedule-*.yml.terraform-init.yml creates the cluster via terraform and outputs k8s-cluster-name. When modifying:
cloud-provider, region, cluster-version, instance-type.terraform-destroy.yml is usually orchestrated to run with always().test-args can be freely combined:
test-args: "--service-version 8.4.3 ${{ inputs.ARGS }}"
test-args: "--service-version 4.2 --topology distribution --replicas 3 ${{ inputs.ARGS }}"
utils.sh --type 36 aggregates test results (used in kubeblocks-e2e-test.yml send-message jobs). See kubeblocks-e2e-workflow for details.
inputs.TEST_TYPE filtering. The if condition uses contains(..., '<engine>|') and
endsWith(..., '<engine>'); for sub-variants (mongodb + mongodb-shard) both must be listed.needs.kbcli-test.yml fails with missing inputskbcli-test.yml must also be added to all callers,
including kubeblocks-e2e-test.yml, kbcli-test-k8s.yml, and cloud-e2e-*.yml.needs matches the actual test job ID.outputs.test-result and outputs.test-result-report.utils.sh --type 36 receives the expected TEST_RESULT format.terraform-init.ymlCLOUD_PROVIDER, REGION, CLUSTER_VERSION, and INSTANCE_TYPE inputs.terraform-destroy-k8s runs before send-messagesend-message-* job to terraform-destroy-k8s.needs.if: always(), so it will run as soon as all explicit needs complete,
including failed jobs.kubeblocks-e2e-test.ymlutils.sh usageGitHub repository automation: PR label, PR size label, PR milestone, issue milestone, auto cherry-pick, comment issue, GitHub Actions runner enable/delete, actions cache cleanup, feishu-message, webhook_utils.sh Use when changing label rules, milestone policies, cherry-pick targets, runner management, or bot message formats.
Cross-registry image mirroring and copy: skopeo sync, skopeo copy, docker.io, apecloud-registry.cn-zhangjiakou.cr.aliyuncs.com, infracreate-registry.cn-zhangjiakou.cr.aliyuncs.com, ECR, set_registry_and_repo.sh. Use when syncing images to Aliyun, copying images between registries, retagging images, or preparing offline image packages.
Container image build, check, push, and manifest management: release-image.yml, buildx, multi-arch amd64 arm64 manifest, helm_image_check.sh, manifests_image_check.sh, skopeo sync, image cache, push_images.sh. Use when image release fails, a new engine image is added, manifest image check fails, or amd64/arm64 platform support is missing.
Handle auxiliary CI/CD workflows in apecloud-cd: workflow trigger, database migration, OpenShift preflight, Red Hat certification. Use for migrate-database.yml, openshift-preflight.yml, or other workflows that do not fit into chart, image, e2e, or release domains.
Helm chart release workflow: helm package, chart-releaser cr, gh-pages index update, Jihulab chart upload, update_chart_values.sh, update_chart_annotations.sh, bump chart version. Use when packaging charts, publishing to helm-charts release, fixing chart index failures, or injecting release metadata into values.yaml and Chart.yaml.
ApeCloud / KubeBlocks Cloud multi-cluster deployment and cleanup: deploy-cloud.yml, cloud-cluster-clear.yml, IDC cleanup, terraform-init.yml, terraform-destroy.yml. Use when deploying to cloud environments, cleaning up IDC/IDC1/IDC2/IDC4 clusters, or debugging deployment failures.