英文:
Mocking errors with client-go fake client
问题
我正在使用client-go（用于Go的Kubernetes客户端）从我的集群中以编程方式检索和更新一些密钥。在这个过程中，我需要对我的代码进行单元测试，经过一些调查，我发现了client-go的fake客户端。然而，我还没有能够模拟错误。我按照this issue中的说明进行了操作，但没有成功。
这是我的业务逻辑：
func (g goClientRefresher) RefreshNamespace(ctx context.Context, namespace string) (err error, warnings bool) {
client := g.kubeClient.CoreV1().Secrets(namespace)
secrets, err := client.List(ctx, metav1.ListOptions{LabelSelector: "mutated-by=confidant"})
if err != nil {
return fmt.Errorf("无法从集群中获取密钥：%w", err), false
}
for _, secret := range secrets.Items {
// 在这里进行业务逻辑处理
}
return nil, warnings
}
以及测试代码：
func TestWhenItsNotPossibleToFetchTheSecrets_ThenAnErrorIsReturned(t *testing.T) {
kubeClient := getKubeClient()
kubeClient.CoreV1().(*fakecorev1.FakeCoreV1).
PrependReactor("list", "secret", func(action testingk8s.Action) (handled bool, ret runtime.Object, err error) {
return true, &v1.SecretList{}, errors.New("无法列出密钥")
})
r := getRefresher(kubeClient)
err, warnings := r.RefreshNamespace(context.Background(), "target-ns")
require.Error(t, err, "应该会出现错误")
}
然而，当我运行测试时，我得到了一个nil错误。我做错了什么吗？
英文:
I'm using client-go (the k8s client for go) to programmatically retrieve and update some secrets from my cluster. While doing this, I'm facing the need of unit-testing my code, and after some investigation I stumbled upon client-go's fake client. However, I haven't been able to mock errors yet. I've followed the instructions from this issue, but without any success.
Here you have my business logic:
func (g goClientRefresher) RefreshNamespace(ctx context.Context, namespace string) (err error, warnings bool) {
client := g.kubeClient.CoreV1().Secrets(namespace)
secrets, err := client.List(ctx, metav1.ListOptions{LabelSelector: "mutated-by=confidant"})
if err != nil {
return fmt.Errorf("unable to fetch secrets from cluster: %w", err), false
}
for _, secret := range secrets.Items {
// business logic here
}
return nil, warnings
}
And the test:
func TestWhenItsNotPossibleToFetchTheSecrets_ThenAnErrorIsReturned(t *testing.T) {
kubeClient := getKubeClient()
kubeClient.CoreV1().(*fakecorev1.FakeCoreV1).
PrependReactor("list", "secret", func(action testingk8s.Action) (handled bool, ret runtime.Object, err error) {
return true, &v1.SecretList{}, errors.New("error listing secrets")
})
r := getRefresher(kubeClient)
err, warnings := r.RefreshNamespace(context.Background(), "target-ns")
require.Error(t, err, "an error should have been raised")
}
However, when I run the test I'm getting a nil error. Am I doing something wrong?
答案1
得分: 1
我终于找到错误了...它在反应堆函数的资源名称中，我写成了secret，而应该是复数形式的secrets... :facepalm:。所以这是代码的正确版本：
func TestWhenItsNotPossibleToFetchTheSecrets_ThenAnErrorIsReturned(t *testing.T) {
kubeClient := getKubeClient()
kubeClient.CoreV1().(*fakecorev1.FakeCoreV1).
PrependReactor("list", "secrets", func(action testingk8s.Action) (handled bool, ret runtime.Object, err error) {
return true, &v1.SecretList{}, errors.New("error listing secrets")
})
// ...
}
英文:
I've finally found the error... it is in the resource name of the reactor function, I had secret and it should be the plural secrets instead... :facepalm:. So this is the correct version of the code:
func TestWhenItsNotPossibleToFetchTheSecrets_ThenAnErrorIsReturned(t *testing.T) {
kubeClient := getKubeClient()
kubeClient.CoreV1().(*fakecorev1.FakeCoreV1).
PrependReactor("list", "secrets", func(action testingk8s.Action) (handled bool, ret runtime.Object, err error) {
return true, &v1.SecretList{}, errors.New("error listing secrets")
})
// ...
}
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论