🌱 test: e2e: make managed suite more robust to errors with Eventually()

test/e2e/data/e2e_eks_conf.yaml

@@ -137,6 +137,7 @@ variables:
   GC_WORKLOAD: "../../data/gcworkload.yaml"
 
 intervals:
+  default/wait-client-request: ["5m", "5s"]
   default/wait-cluster: ["40m", "10s"]
   default/wait-control-plane: ["35m", "10s"]
   default/wait-worker-nodes: ["30m", "10s"]

test/e2e/suites/managed/addon.go

@@ -62,8 +62,9 @@ func CheckAddonExistsSpec(ctx context.Context, inputGetter func() CheckAddonExis
 
 	By(fmt.Sprintf("Getting control plane: %s", controlPlaneName))
 	controlPlane := &ekscontrolplanev1.AWSManagedControlPlane{}
-	err := mgmtClient.Get(ctx, crclient.ObjectKey{Namespace: input.Namespace.Name, Name: controlPlaneName}, controlPlane)
-	Expect(err).ToNot(HaveOccurred())
+	Eventually(func() error {
+		return mgmtClient.Get(ctx, crclient.ObjectKey{Namespace: input.Namespace.Name, Name: controlPlaneName}, controlPlane)
+	}, input.E2EConfig.GetIntervals("", "wait-client-request")...).Should(Succeed(), "eventually failed trying to get the AWSManagedControlPlane")
 
 	By(fmt.Sprintf("Checking EKS addon %s is installed on cluster %s and is active", input.AddonName, input.ClusterName))
 	waitForEKSAddonToHaveStatus(waitForEKSAddonToHaveStatusInput{

test/e2e/suites/managed/aws_node_env.go

@@ -21,6 +21,7 @@ package managed
 
 import (
 	"context"
+	"errors"
 	"fmt"
 
 	"github.com/aws/aws-sdk-go/aws/client"
@@ -57,27 +58,34 @@ func CheckAwsNodeEnvVarsSet(ctx context.Context, inputGetter func() UpdateAwsNod
 
 	By(fmt.Sprintf("Getting control plane: %s", controlPlaneName))
 	controlPlane := &ekscontrolplanev1.AWSManagedControlPlane{}
-	err := mgmtClient.Get(ctx, crclient.ObjectKey{Namespace: input.Namespace.Name, Name: controlPlaneName}, controlPlane)
-	Expect(err).ToNot(HaveOccurred())
+	Eventually(func() error {
+		return mgmtClient.Get(ctx, crclient.ObjectKey{Namespace: input.Namespace.Name, Name: controlPlaneName}, controlPlane)
+	}, input.E2EConfig.GetIntervals("", "wait-client-request")...).Should(Succeed(), "eventually failed trying to get the AWSManagedControlPlane")
 
 	By(fmt.Sprintf("Checking environment variables are set on AWSManagedControlPlane: %s", controlPlaneName))
 	Expect(controlPlane.Spec.VpcCni.Env).NotTo(BeNil())
 	Expect(len(controlPlane.Spec.VpcCni.Env)).Should(BeNumerically(">", 1))
 
 	By("Checking if aws-node has been updated with the defined environment variables on the workload cluster")
 	daemonSet := &appsv1.DaemonSet{}
-
 	clusterClient := input.BootstrapClusterProxy.GetWorkloadCluster(ctx, input.Namespace.Name, input.ClusterName).GetClient()
-	err = clusterClient.Get(ctx, crclient.ObjectKey{Namespace: "kube-system", Name: "aws-node"}, daemonSet)
-	Expect(err).ToNot(HaveOccurred())
-
-	for _, container := range daemonSet.Spec.Template.Spec.Containers {
-		if container.Name == "aws-node" {
-			Expect(matchEnvVar(container.Env, corev1.EnvVar{Name: "FOO", Value: "BAR"})).Should(BeTrue())
-			Expect(matchEnvVar(container.Env, corev1.EnvVar{Name: "ENABLE_PREFIX_DELEGATION", Value: "true"})).Should(BeTrue())
-			break
+
+	Eventually(func() error {
+		if err := clusterClient.Get(ctx, crclient.ObjectKey{Namespace: "kube-system", Name: "aws-node"}, daemonSet); err != nil {
+			return fmt.Errorf("unable to get aws-node: %w", err)
 		}
-	}
+
+		for _, container := range daemonSet.Spec.Template.Spec.Containers {
+			if container.Name == "aws-node" {
+				if matchEnvVar(container.Env, corev1.EnvVar{Name: "FOO", Value: "BAR"}) &&
+					matchEnvVar(container.Env, corev1.EnvVar{Name: "ENABLE_PREFIX_DELEGATION", Value: "true"}) {
+					return nil
+				}
+			}
+		}
+
+		return errors.New("unable to find the expected environment variables on the aws-node DaemonSet's container")
+	}, input.E2EConfig.GetIntervals("", "wait-client-request")...).Should(Succeed(), "should have been able to find the expected aws-node DaemonSet")
 }
 
 func matchEnvVar(s []corev1.EnvVar, ev corev1.EnvVar) bool {

test/e2e/suites/managed/eks_test.go

@@ -22,7 +22,6 @@ package managed
 import (
 	"context"
 	"fmt"
-	"time"
 
 	"github.com/onsi/ginkgo/v2"
 	. "github.com/onsi/gomega"
@@ -75,22 +74,15 @@ var _ = ginkgo.Describe("[managed] [general] EKS cluster tests", func() {
 		})
 
 		ginkgo.By("should set environment variables on the aws-node daemonset")
-		Eventually(func() error {
-			defer ginkgo.GinkgoRecover()
-			CheckAwsNodeEnvVarsSet(ctx, func() UpdateAwsNodeVersionSpecInput {
-				return UpdateAwsNodeVersionSpecInput{
-					E2EConfig:             e2eCtx.E2EConfig,
-					BootstrapClusterProxy: e2eCtx.Environment.BootstrapClusterProxy,
-					AWSSession:            e2eCtx.BootstrapUserAWSSession,
-					Namespace:             namespace,
-					ClusterName:           clusterName,
-				}
-			})
-			return nil
-		}).WithTimeout(5*time.Minute).WithPolling(10*time.Second).WithContext(ctx).Should(
-			Succeed(),
-			"Failed to verify AWS Node environment variables after 5 minutes of retries",
-		)
+		CheckAwsNodeEnvVarsSet(ctx, func() UpdateAwsNodeVersionSpecInput {
+			return UpdateAwsNodeVersionSpecInput{
+				E2EConfig:             e2eCtx.E2EConfig,
+				BootstrapClusterProxy: e2eCtx.Environment.BootstrapClusterProxy,
+				AWSSession:            e2eCtx.BootstrapUserAWSSession,
+				Namespace:             namespace,
+				ClusterName:           clusterName,
+			}
+		})
 
 		ginkgo.By("should have the VPC CNI installed")
 		CheckAddonExistsSpec(ctx, func() CheckAddonExistsSpecInput {

test/e2e/suites/managed/helpers.go

@@ -32,6 +32,7 @@ import (
 	. "github.com/onsi/gomega"
 	corev1 "k8s.io/api/core/v1"
 	apimachinerytypes "k8s.io/apimachinery/pkg/types"
+	"k8s.io/utils/ptr"
 	crclient "sigs.k8s.io/controller-runtime/pkg/client"
 
 	infrav1 "sigs.k8s.io/cluster-api-provider-aws/v2/api/v1beta2"
@@ -51,6 +52,11 @@ const (
 	EKSControlPlaneOnlyLegacyFlavor                   = "eks-control-plane-only-legacy"
 )
 
+const (
+	clientRequestTimeout       = 2 * time.Minute
+	clientRequestCheckInterval = 5 * time.Second
+)
+
 type DefaultConfigClusterFn func(clusterName, namespace string) clusterctl.ConfigClusterInput
 
 func getEKSClusterName(namespace, clusterName string) string {
@@ -74,14 +80,19 @@ func getASGName(clusterName string) string {
 }
 
 func verifyClusterActiveAndOwned(eksClusterName string, sess client.ConfigProvider) {
-	cluster, err := getEKSCluster(eksClusterName, sess)
-	Expect(err).NotTo(HaveOccurred())
+	var (
+		cluster *eks.Cluster
+		err     error
+	)
+	Eventually(func() error {
+		cluster, err = getEKSCluster(eksClusterName, sess)
+		return err
+	}, clientRequestTimeout, clientRequestCheckInterval).Should(Succeed(), fmt.Sprintf("eventually failed trying to get EKS Cluster %q", eksClusterName))
 
 	tagName := infrav1.ClusterTagKey(eksClusterName)
 	tagValue, ok := cluster.Tags[tagName]
 	Expect(ok).To(BeTrue(), "expecting the cluster owned tag to exist")
 	Expect(*tagValue).To(BeEquivalentTo(string(infrav1.ResourceLifecycleOwned)))
-
 	Expect(*cluster.Status).To(BeEquivalentTo(eks.ClusterStatusActive))
 }
 
@@ -102,6 +113,7 @@ func getEKSClusterAddon(eksClusterName, addonName string, sess client.ConfigProv
 		AddonName:   &addonName,
 		ClusterName: &eksClusterName,
 	}
+
 	describeOutput, err := eksClient.DescribeAddon(describeInput)
 	if err != nil {
 		return nil, fmt.Errorf("describing eks addon %s: %w", addonName, err)
@@ -112,16 +124,16 @@ func getEKSClusterAddon(eksClusterName, addonName string, sess client.ConfigProv
 
 func verifySecretExists(ctx context.Context, secretName, namespace string, k8sclient crclient.Client) {
 	secret := &corev1.Secret{}
-	err := k8sclient.Get(ctx, apimachinerytypes.NamespacedName{Name: secretName, Namespace: namespace}, secret)
-
-	Expect(err).ShouldNot(HaveOccurred())
+	Eventually(func() error {
+		return k8sclient.Get(ctx, apimachinerytypes.NamespacedName{Name: secretName, Namespace: namespace}, secret)
+	}, clientRequestTimeout, clientRequestCheckInterval).Should(Succeed(), fmt.Sprintf("eventually failed trying to verify Secret %q exists", secretName))
 }
 
 func verifyConfigMapExists(ctx context.Context, name, namespace string, k8sclient crclient.Client) {
 	cm := &corev1.ConfigMap{}
 	Eventually(func() error {
 		return k8sclient.Get(ctx, apimachinerytypes.NamespacedName{Name: name, Namespace: namespace}, cm)
-	}, 2*time.Minute, 5*time.Second).Should(Succeed())
+	}, clientRequestTimeout, clientRequestCheckInterval).Should(Succeed(), fmt.Sprintf("eventually failed trying to verify ConfigMap %q exists", name))
 }
 
 func VerifyRoleExistsAndOwned(roleName string, eksClusterName string, checkOwned bool, sess client.ConfigProvider) {
@@ -130,8 +142,15 @@ func VerifyRoleExistsAndOwned(roleName string, eksClusterName string, checkOwned
 		RoleName: aws.String(roleName),
 	}
 
-	output, err := iamClient.GetRole(input)
-	Expect(err).ShouldNot(HaveOccurred())
+	var (
+		output *iam.GetRoleOutput
+		err    error
+	)
+
+	Eventually(func() error {
+		output, err = iamClient.GetRole(input)
+		return err
+	}, clientRequestTimeout, clientRequestCheckInterval).Should(Succeed(), fmt.Sprintf("eventually failed trying to get IAM Role %q", roleName))
 
 	if checkOwned {
 		found := false
@@ -152,9 +171,24 @@ func verifyManagedNodeGroup(eksClusterName, nodeGroupName string, checkOwned boo
 		ClusterName:   aws.String(eksClusterName),
 		NodegroupName: aws.String(nodeGroupName),
 	}
-	result, err := eksClient.DescribeNodegroup(input)
-	Expect(err).NotTo(HaveOccurred())
-	Expect(*result.Nodegroup.Status).To(BeEquivalentTo(eks.NodegroupStatusActive))
+	var (
+		result *eks.DescribeNodegroupOutput
+		err    error
+	)
+
+	Eventually(func() error {
+		result, err = eksClient.DescribeNodegroup(input)
+		if err != nil {
+			return fmt.Errorf("error describing nodegroup: %w", err)
+		}
+
+		nodeGroupStatus := ptr.Deref(result.Nodegroup.Status, "")
+		if nodeGroupStatus != eks.NodegroupStatusActive {
+			return fmt.Errorf("expected nodegroup.Status to be %q, was %q instead", eks.NodegroupStatusActive, nodeGroupStatus)
+		}
+
+		return nil
+	}, clientRequestTimeout, clientRequestCheckInterval).Should(Succeed(), "eventually failed trying to describe EKS Node group")
 
 	if checkOwned {
 		tagName := infrav1.ClusterAWSCloudProviderTagKey(eksClusterName)
@@ -172,8 +206,16 @@ func verifyASG(eksClusterName, asgName string, checkOwned bool, sess client.Conf
 		},
 	}
 
-	result, err := asgClient.DescribeAutoScalingGroups(input)
-	Expect(err).NotTo(HaveOccurred())
+	var (
+		result *autoscaling.DescribeAutoScalingGroupsOutput
+		err    error
+	)
+
+	Eventually(func() error {
+		result, err = asgClient.DescribeAutoScalingGroups(input)
+		return err
+	}, clientRequestTimeout, clientRequestCheckInterval).Should(Succeed())
+
 	for _, instance := range result.AutoScalingGroups[0].Instances {
 		Expect(*instance.LifecycleState).To(Equal("InService"), "expecting the instance in service")
 	}