test(kubernetes): Test TailContainer happy path

runtime/kubernetes/container_test.go

@@ -208,17 +208,54 @@ func TestKubernetes_SetupContainer(t *testing.T) {
 	}
 }
 
-// TODO: implement this once they resolve the bug
-//
-// https://github.com/kubernetes/kubernetes/issues/84203
 func TestKubernetes_TailContainer(t *testing.T) {
-	// Unfortunately, we can't implement this test using
-	// the native Kubernetes fake. This is because there
-	// is a bug in that code where an "empty" request is
-	// always returned when calling the GetLogs function.
+	// Unfortunately, we can't test failures using the native Kubernetes fake.
+	// k8s.client-go v0.19.0 added a mock GetLogs() response so that
+	// it no longer panics with an "empty" request, but now it always returns
+	// a successful response with Body: "fake logs".
 	//
 	// https://github.com/kubernetes/kubernetes/issues/84203
-	// fixed in k8s.io/client-go v0.19.0; we already have v0.22.2
+	// https://github.com/kubernetes/kubernetes/pulls/91485
+	//
+	// setup types
+	_engine, err := NewMock(_pod)
+	if err != nil {
+		t.Errorf("unable to create runtime engine: %v", err)
+	}
+
+	// setup tests
+	tests := []struct {
+		failure   bool
+		container *pipeline.Container
+	}{
+		{
+			failure:   false,
+			container: _container,
+		},
+		// We cannot test failures, because the mock GetLogs() always
+		// returns a successful response with logs body: "fake logs"
+		//{
+		//	failure:   true,
+		//	container: new(pipeline.Container),
+		//},
+	}
+
+	// run tests
+	for _, test := range tests {
+		_, err = _engine.TailContainer(context.Background(), test.container)
+
+		if test.failure {
+			if err == nil {
+				t.Errorf("TailContainer should have returned err")
+			}
+
+			continue
+		}
+
+		if err != nil {
+			t.Errorf("TailContainer returned err: %v", err)
+		}
+	}
 }
 
 func TestKubernetes_WaitContainer(t *testing.T) {