Merge pull request kubernetes#128946 from jackfrancis/SetPVCVACName-eventually-gomega

test: don't panic during an Eventually retry loop

Author: k8s-ci-robot

test/e2e/instrumentation/metrics.go

@@ -44,10 +44,7 @@ var _ = common.SIGDescribe("Metrics", func() {
 		ec = f.KubemarkExternalClusterClientSet
 		gomega.Eventually(ctx, func() error {
 			grabber, err = e2emetrics.NewMetricsGrabber(ctx, c, ec, f.ClientConfig(), true, true, true, true, true, true)
-			if err != nil {
-				framework.ExpectNoError(err, "failed to create metrics grabber")
-			}
-			return nil
+			return err
 		}, 5*time.Minute, 10*time.Second).Should(gomega.BeNil())
 	})
 

test/e2e/node/gpu.go

@@ -18,6 +18,7 @@ package node
 
 import (
 	"context"
+	"fmt"
 	"os"
 	"regexp"
 	"time"
@@ -292,10 +293,12 @@ func SetupEnvironmentAndSkipIfNeeded(ctx context.Context, f *framework.Framework
 	}
 }
 
-func areGPUsAvailableOnAllSchedulableNodes(ctx context.Context, clientSet clientset.Interface) bool {
+func areGPUsAvailableOnAllSchedulableNodes(ctx context.Context, clientSet clientset.Interface) error {
 	framework.Logf("Getting list of Nodes from API server")
 	nodeList, err := clientSet.CoreV1().Nodes().List(ctx, metav1.ListOptions{})
-	framework.ExpectNoError(err, "getting node list")
+	if err != nil {
+		return fmt.Errorf("unexpected error getting node list: %w", err)
+	}
 	for _, node := range nodeList.Items {
 		if node.Spec.Unschedulable {
 			continue
@@ -305,12 +308,11 @@ func areGPUsAvailableOnAllSchedulableNodes(ctx context.Context, clientSet client
 		}
 		framework.Logf("gpuResourceName %s", e2egpu.NVIDIAGPUResourceName)
 		if val, ok := node.Status.Capacity[e2egpu.NVIDIAGPUResourceName]; !ok || val.Value() == 0 {
-			framework.Logf("Nvidia GPUs not available on Node: %q", node.Name)
-			return false
+			return fmt.Errorf("nvidia GPUs not available on Node: %q", node.Name)
 		}
 	}
 	framework.Logf("Nvidia GPUs exist on all schedulable nodes")
-	return true
+	return nil
 }
 
 func logOSImages(ctx context.Context, f *framework.Framework) {
@@ -386,9 +388,9 @@ func waitForGPUs(ctx context.Context, f *framework.Framework, namespace, name st
 
 	// Wait for Nvidia GPUs to be available on nodes
 	framework.Logf("Waiting for drivers to be installed and GPUs to be available in Node Capacity...")
-	gomega.Eventually(ctx, func(ctx context.Context) bool {
+	gomega.Eventually(ctx, func(ctx context.Context) error {
 		return areGPUsAvailableOnAllSchedulableNodes(ctx, f.ClientSet)
-	}, driverInstallTimeout, time.Second).Should(gomega.BeTrueBecause("expected GPU resources to be available within the timout"))
+	}, driverInstallTimeout, time.Second).Should(gomega.Succeed())
 }
 
 // StartJob starts a simple CUDA job that requests gpu and the specified number of completions

test/e2e/storage/testsuites/volume_modify.go

@@ -398,13 +398,13 @@ func SetPVCVACName(ctx context.Context, origPVC *v1.PersistentVolumeClaim, name
 	pvcName := origPVC.Name
 	var patchedPVC *v1.PersistentVolumeClaim
 
-	gomega.Eventually(ctx, func(g gomega.Gomega) {
+	gomega.Eventually(ctx, func() error {
 		var err error
 		patch := []map[string]interface{}{{"op": "replace", "path": "/spec/volumeAttributesClassName", "value": name}}
 		patchBytes, _ := json.Marshal(patch)
 
 		patchedPVC, err = c.CoreV1().PersistentVolumeClaims(origPVC.Namespace).Patch(ctx, pvcName, types.JSONPatchType, patchBytes, metav1.PatchOptions{})
-		framework.ExpectNoError(err, "While patching PVC to add VAC name")
+		return err
 	}, timeout, modifyPollInterval).Should(gomega.Succeed())
 
 	return patchedPVC

test/e2e/windows/eviction.go

@@ -18,6 +18,7 @@ package windows
 
 import (
 	"context"
+	"fmt"
 	"strconv"
 	"strings"
 	"time"
@@ -176,18 +177,20 @@ var _ = sigDescribe(feature.Windows, "Eviction", framework.WithSerial(), framewo
 		framework.ExpectNoError(err)
 
 		framework.Logf("Waiting for pod2 to get evicted")
-		gomega.Eventually(ctx, func() bool {
+		gomega.Eventually(ctx, func() error {
 			eventList, err := f.ClientSet.CoreV1().Events(f.Namespace.Name).List(ctx, metav1.ListOptions{})
-			framework.ExpectNoError(err)
+			if err != nil {
+				return fmt.Errorf("error getting events: %w", err)
+			}
 			for _, e := range eventList.Items {
 				// Look for an event that shows FailedScheduling
 				if e.Type == "Warning" && e.Reason == "Evicted" && strings.Contains(e.Message, "pod2") {
 					framework.Logf("Found %+v event with message %+v", e.Reason, e.Message)
-					return true
+					return nil
 				}
 			}
-			return false
-		}, 10*time.Minute, 10*time.Second).Should(gomega.BeTrueBecause("Eviction Event was not found"))
+			return fmt.Errorf("did not find any FailedScheduling event for pod %s", pod2.ObjectMeta.Name)
+		}, 10*time.Minute, 10*time.Second).Should(gomega.Succeed())
 
 		ginkgo.By("Waiting for node.kubernetes.io/memory-pressure taint to be removed")
 		// ensure e2e test framework catches the memory-pressure taint

test/e2e/windows/memory_limits.go

@@ -19,6 +19,7 @@ package windows
 import (
 	"context"
 	"encoding/json"
+	"fmt"
 	"time"
 
 	kubeletconfigv1beta1 "k8s.io/kubelet/config/v1beta1"
@@ -162,18 +163,20 @@ func overrideAllocatableMemoryTest(ctx context.Context, f *framework.Framework,
 	framework.Logf("Ensuring that pod %s fails to schedule", podName)
 	failurePod, err = f.ClientSet.CoreV1().Pods(f.Namespace.Name).Create(ctx, failurePod, metav1.CreateOptions{})
 	framework.ExpectNoError(err)
-	gomega.Eventually(ctx, func() bool {
+	gomega.Eventually(ctx, func() error {
 		eventList, err := f.ClientSet.CoreV1().Events(f.Namespace.Name).List(ctx, metav1.ListOptions{})
-		framework.ExpectNoError(err)
+		if err != nil {
+			return fmt.Errorf("error getting events: %w", err)
+		}
 		for _, e := range eventList.Items {
 			// Look for an event that shows FailedScheduling
 			if e.Type == "Warning" && e.Reason == "FailedScheduling" && e.InvolvedObject.Name == failurePod.ObjectMeta.Name {
 				framework.Logf("Found %+v event with message %+v", e.Reason, e.Message)
-				return true
+				return nil
 			}
 		}
-		return false
-	}, 3*time.Minute, 10*time.Second).Should(gomega.BeTrueBecause("Expected %s pod to be failed scheduling", podName))
+		return fmt.Errorf("did not find any FailedScheduling event for pod %s", failurePod.ObjectMeta.Name)
+	}, 3*time.Minute, 10*time.Second).Should(gomega.Succeed())
 }
 
 func getNodeMemory(ctx context.Context, f *framework.Framework, node v1.Node) nodeMemory {

test/e2e_node/util.go

@@ -237,11 +237,16 @@ func waitForKubeletToStart(ctx context.Context, f *framework.Framework) {
 	}, 2*time.Minute, 5*time.Second).Should(gomega.BeTrueBecause("expected kubelet to be in healthy state"))
 
 	// Wait for the Kubelet to be ready.
-	gomega.Eventually(ctx, func(ctx context.Context) bool {
+	gomega.Eventually(ctx, func(ctx context.Context) error {
 		nodes, err := e2enode.TotalReady(ctx, f.ClientSet)
-		framework.ExpectNoError(err)
-		return nodes == 1
-	}, time.Minute, time.Second).Should(gomega.BeTrueBecause("expected kubelet to be in ready state"))
+		if err != nil {
+			return fmt.Errorf("error getting ready nodes: %w", err)
+		}
+		if nodes != 1 {
+			return fmt.Errorf("expected 1 ready node, got %d", nodes)
+		}
+		return nil
+	}, time.Minute, time.Second).Should(gomega.Succeed())
 }
 
 func deleteStateFile(stateFileName string) {