菜单
Run ORC tests (unit tests, linting, and E2E tests). Use after making changes to verify correctness.
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 | testing |
| description | Run ORC tests (unit tests, linting, and E2E tests). Use after making changes to verify correctness. |
| disable-model-invocation | true |
Run unit tests, linting, and E2E tests for ORC controllers.
Before running E2E tests, ensure code compiles and passes linting:
make generate
make lint
make test
E2E tests require E2E_OSCLOUDS environment variable pointing to a clouds.yaml file containing cloud entries for regular and admin credentials. The cloud names are configurable via environment variables:
| Variable | Description | Default |
|---|---|---|
E2E_OSCLOUDS | Path to clouds.yaml | /etc/openstack/clouds.yaml |
E2E_OPENSTACK_CLOUD_NAME | Cloud name for regular credentials | devstack |
E2E_OPENSTACK_ADMIN_CLOUD_NAME | Cloud name for admin credentials | devstack-admin-demo |
If the user did not provide E2E_OSCLOUDS, tell them local E2E testing will be skipped and they should run it manually later or in CI.
If E2E_OSCLOUDS is provided, execute each step in order:
Step 1: Create kind cluster (if not already running)
# Check if cluster exists
kind get clusters
# Create only if no cluster exists
kind create cluster
If a cluster already exists, skip creation and proceed to Step 2.
Step 2: Verify cluster is ready
kubectl get nodes
Ensure node shows Ready status.
Step 3: Install CRDs
kubectl apply -k config/crd --server-side
Step 4: Stop any existing manager, rebuild, and start
# Stop any existing manager to ensure we're running latest code
pkill -f orc-manager || true
# Build and start fresh
go build -o /tmp/orc-manager ./cmd/manager
/tmp/orc-manager -zap-log-level 5 > /tmp/manager.log 2>&1 &
Step 5: Wait for manager to start and verify it's running
sleep 5
ps aux | grep "[o]rc-manager"
If no process found, check /tmp/manager.log for errors.
Step 6: Run E2E tests
Replace /path/to/clouds.yaml with the actual path and <kind> with the controller name:
E2E_OSCLOUDS=/path/to/clouds.yaml E2E_KUTTL_DIR=internal/controllers/<kind>/tests make test-e2e
Step 7: If tests fail, review manager logs
# Search for errors first (logs are verbose at level 5)
grep -i error /tmp/manager.log | tail -50
# Or view more context
tail -500 /tmp/manager.log
Use these logs to diagnose and fix issues, then re-run the tests.
Step 8: Cleanup After tests pass (or when done debugging):
pkill -f "orc-manager" || true
kind delete cluster
rm -f /tmp/manager.log /tmp/orc-manager
Tests are located in internal/controllers/<kind>/tests/:
| Directory | Purpose |
|---|---|
<kind>-create-minimal/ | Create with minimum required fields |
<kind>-create-full/ | Create with all fields |
<kind>-import/ | Import existing resource |
<kind>-import-error/ | Import with no matches |
<kind>-dependency/ | Test dependency waiting and deletion guards |
<kind>-import-dependency/ | Test import with dependency references |
<kind>-update/ | Test mutable field updates |
Review ORC controller code for Kubernetes best practices and ORC conventions. Use after implementing or modifying a controller.
Add a dependency on another ORC resource to a controller. Use when a resource needs to reference or wait for another resource (e.g., Subnet depends on Network).
Create a new ORC controller for an OpenStack resource. Use when adding support for a new OpenStack resource type (e.g., LoadBalancer, FloatingIP).
Write an enhancement proposal for a new ORC feature. Use for significant new features, breaking changes, or cross-cutting architectural changes.
Update an existing ORC controller. Use when adding fields, making fields mutable, adding tag support, or improving error handling.
Draft release notes for a new ORC version. Use when preparing a release to generate changelog and GitHub release body from git history.