e2e: must-gather: fix the setup process

test/e2e/install/install_test.go

@@ -77,7 +77,7 @@ var _ = Describe("[Install] continuousIntegration", Serial, func() {
 	Context("with a running cluster with all the components", func() {
 		It("[test_id:47574][tier0] should perform overall deployment and verify the condition is reported as available", func() {
 			deployer := deploy.NewForPlatform(configuration.Plat)
-			nroObj := deployer.Deploy(context.TODO())
+			nroObj := deployer.Deploy(context.TODO(), configuration.MachineConfigPoolUpdateTimeout)
 			nname := client.ObjectKeyFromObject(nroObj)
 			Expect(nname.Name).ToNot(BeEmpty())
 
@@ -205,7 +205,7 @@ var _ = Describe("[Install] durability", Serial, func() {
 
 		BeforeEach(func() {
 			deployer = deploy.NewForPlatform(configuration.Plat)
-			nroObj = deployer.Deploy(context.TODO())
+			nroObj = deployer.Deploy(context.TODO(), configuration.MachineConfigPoolUpdateTimeout)
 		})
 
 		AfterEach(func() {

test/e2e/must-gather/must_gather_suite_test.go

@@ -25,8 +25,11 @@ import (
 
 	"github.com/k8stopologyawareschedwg/deployer/pkg/deployer/platform"
 
+	nropv1 "github.com/openshift-kni/numaresources-operator/api/numaresourcesoperator/v1"
+	e2eclient "github.com/openshift-kni/numaresources-operator/test/utils/clients"
 	"github.com/openshift-kni/numaresources-operator/test/utils/configuration"
 	"github.com/openshift-kni/numaresources-operator/test/utils/deploy"
+	"github.com/openshift-kni/numaresources-operator/test/utils/objects"
 
 	"github.com/onsi/ginkgo/v2"
 	"github.com/onsi/gomega"
@@ -38,10 +41,14 @@ const (
 
 	defaultMustGatherImage = "quay.io/openshift-kni/numaresources-must-gather"
 	defaultMustGatherTag   = "4.19.999-snapshot"
+
+	nroSchedTimeout = 5 * time.Minute
 )
 
 var (
-	deployment deploy.NroDeploymentWithSched
+	deployment deploy.Deployer
+
+	nroSchedObj *nropv1.NUMAResourcesScheduler
 
 	mustGatherImage string
 	mustGatherTag   string
@@ -62,17 +69,23 @@ var _ = ginkgo.BeforeSuite(func() {
 	mustGatherTag = getStringValueFromEnv(envVarMustGatherTag, defaultMustGatherTag)
 	ginkgo.By(fmt.Sprintf("Using must-gather image %q tag %q", mustGatherImage, mustGatherTag))
 
+	ctx := context.Background()
+
 	if _, ok := os.LookupEnv("E2E_NROP_INFRA_SETUP_SKIP"); ok {
 		ginkgo.By("Fetching up cluster data")
 
-		var err error
-		deployment, err = deploy.GetDeploymentWithSched(context.TODO())
-		gomega.Expect(err).ToNot(gomega.HaveOccurred())
+		// assume cluster is set up correctly, so just fetch what we have already;
+		// fail loudly if we can't get, this means the assumption was wrong
+		nroSchedObj = &nropv1.NUMAResourcesScheduler{}
+		gomega.Expect(e2eclient.Client.Get(ctx, objects.NROSchedObjectKey(), nroSchedObj)).To(gomega.Succeed())
 		return
 	}
+
 	ginkgo.By("Setting up the cluster")
-	deployment.Deploy(context.TODO())
-	deployment.NroSchedObj = deploy.DeployNROScheduler()
+
+	deployment = deploy.NewForPlatform(configuration.Plat)
+	_ = deployment.Deploy(ctx, configuration.MachineConfigPoolUpdateTimeout) // we don't care about the nrop instance
+	nroSchedObj = deploy.DeployNROScheduler(ctx, nroSchedTimeout)
 })
 
 var _ = ginkgo.AfterSuite(func() {
@@ -81,8 +94,9 @@ var _ = ginkgo.AfterSuite(func() {
 		return
 	}
 	ginkgo.By("tearing down the cluster")
-	deploy.TeardownNROScheduler(deployment.NroSchedObj, 5*time.Minute)
-	deployment.Teardown(context.TODO(), 5*time.Minute)
+	ctx := context.Background()
+	deploy.TeardownNROScheduler(ctx, nroSchedObj, nroSchedTimeout)
+	deployment.Teardown(ctx, 5*time.Minute)
 })
 
 func getStringValueFromEnv(envVar, fallback string) string {

test/e2e/must-gather/must_gather_test.go

@@ -45,6 +45,8 @@ var _ = Describe("[must-gather] NRO data collected", func() {
 		var destDir string
 
 		BeforeEach(func() {
+			Expect(nroSchedObj).ToNot(BeNil(), "missing scheduler object reference")
+
 			var err error
 			destDir, err = os.MkdirTemp("", "*-e2e-data")
 			Expect(err).ToNot(HaveOccurred())
@@ -106,7 +108,7 @@ var _ = Describe("[must-gather] NRO data collected", func() {
 
 				By("Looking for resources instances")
 				nropInstanceFileName := fmt.Sprintf("%s.yaml", filepath.Join("cluster-scoped-resources/nodetopology.openshift.io/numaresourcesoperators", objects.NROObjectKey().Name))
-				nroschedInstanceFileName := fmt.Sprintf("%s.yaml", filepath.Join("cluster-scoped-resources/nodetopology.openshift.io/numaresourcesschedulers", deployment.NroSchedObj.Name))
+				nroschedInstanceFileName := fmt.Sprintf("%s.yaml", filepath.Join("cluster-scoped-resources/nodetopology.openshift.io/numaresourcesschedulers", nroSchedObj.Name))
 
 				collectedMCPs, err := getMachineConfigPools(filepath.Join(mgContentFolder, "cluster-scoped-resources/machineconfiguration.openshift.io/machineconfigpools"))
 				Expect(err).ToNot(HaveOccurred())

test/utils/deploy/deploy.go

@@ -43,20 +43,19 @@ import (
 	"github.com/openshift-kni/numaresources-operator/test/utils/objects"
 )
 
+const (
+	NROSchedulerPollingInterval = 10 * time.Second
+)
+
 type Deployer interface {
 	// Deploy deploys NUMAResourcesOperator object create other dependencies
 	// per the platform that implements it
-	Deploy(ctx context.Context) *nropv1.NUMAResourcesOperator
+	Deploy(ctx context.Context, timeout time.Duration) *nropv1.NUMAResourcesOperator
 	// Teardown Teardowns NUMAResourcesOperator object delete other dependencies
 	// per the platform that implements it
 	Teardown(ctx context.Context, timeout time.Duration)
 }
 
-type NroDeploymentWithSched struct {
-	Deployer
-	NroSchedObj *nropv1.NUMAResourcesScheduler
-}
-
 func NewForPlatform(plat platform.Platform) Deployer {
 	switch plat {
 	case platform.OpenShift:
@@ -70,19 +69,6 @@ func NewForPlatform(plat platform.Platform) Deployer {
 	}
 }
 
-func GetDeploymentWithSched(ctx context.Context) (NroDeploymentWithSched, error) {
-	sd := NroDeploymentWithSched{}
-	nroSchedKey := objects.NROSchedObjectKey()
-	nroSchedObj := nropv1.NUMAResourcesScheduler{}
-	err := e2eclient.Client.Get(ctx, nroSchedKey, &nroSchedObj)
-	if err != nil {
-		return sd, err
-	}
-	sd.NroSchedObj = &nroSchedObj
-
-	return sd, nil
-}
-
 func WaitForMCPUpdatedAfterNRODeleted(nroObj *nropv1.NUMAResourcesOperator) {
 	GinkgoHelper()
 
@@ -122,20 +108,16 @@ func isMachineConfigPoolsUpdatedAfterDeletion(nro *nropv1.NUMAResourcesOperator)
 // or a timeout happens (5 min right now).
 //
 // see: `TestNROScheduler` to see the specific object characteristics.
-func DeployNROScheduler() *nropv1.NUMAResourcesScheduler {
+func DeployNROScheduler(ctx context.Context, timeout time.Duration) *nropv1.NUMAResourcesScheduler {
 	GinkgoHelper()
 
 	nroSchedObj := objects.TestNROScheduler()
 
-	err := e2eclient.Client.Create(context.TODO(), nroSchedObj)
-	Expect(err).NotTo(HaveOccurred())
-
-	err = e2eclient.Client.Get(context.TODO(), client.ObjectKeyFromObject(nroSchedObj), nroSchedObj)
-	Expect(err).NotTo(HaveOccurred())
-
+	Expect(e2eclient.Client.Create(ctx, nroSchedObj)).To(Succeed())
+	Expect(e2eclient.Client.Get(ctx, client.ObjectKeyFromObject(nroSchedObj), nroSchedObj)).To(Succeed())
 	Eventually(func() bool {
 		updatedNROObj := &nropv1.NUMAResourcesScheduler{}
-		err := e2eclient.Client.Get(context.TODO(), client.ObjectKeyFromObject(nroSchedObj), updatedNROObj)
+		err := e2eclient.Client.Get(ctx, client.ObjectKeyFromObject(nroSchedObj), updatedNROObj)
 		if err != nil {
 			klog.Warningf("failed to get the NRO Scheduler resource: %v", err)
 			return false
@@ -151,20 +133,15 @@ func DeployNROScheduler() *nropv1.NUMAResourcesScheduler {
 		klog.Infof("condition: %v", cond)
 
 		return cond.Status == metav1.ConditionTrue
-	}).WithTimeout(5*time.Minute).WithPolling(10*time.Second).Should(BeTrue(), "NRO Scheduler condition did not become available")
+	}).WithTimeout(timeout).WithPolling(NROSchedulerPollingInterval).Should(BeTrue(), "NRO Scheduler condition did not become available")
 	return nroSchedObj
 }
 
-func TeardownNROScheduler(nroSched *nropv1.NUMAResourcesScheduler, timeout time.Duration) {
+func TeardownNROScheduler(ctx context.Context, nroSched *nropv1.NUMAResourcesScheduler, timeout time.Duration) {
 	GinkgoHelper()
-
-	if nroSched != nil {
-		err := e2eclient.Client.Delete(context.TODO(), nroSched)
-		Expect(err).ToNot(HaveOccurred())
-
-		err = wait.With(e2eclient.Client).Interval(10*time.Second).Timeout(timeout).ForNUMAResourcesSchedulerDeleted(context.TODO(), nroSched)
-		Expect(err).ToNot(HaveOccurred(), "NROScheduler %q failed to be deleted", nroSched.Name)
-	}
+	Expect(nroSched).ToNot(BeNil())
+	Expect(e2eclient.Client.Delete(ctx, nroSched)).To(Succeed())
+	Expect(wait.With(e2eclient.Client).Interval(NROSchedulerPollingInterval).Timeout(timeout).ForNUMAResourcesSchedulerDeleted(ctx, nroSched)).To(Succeed(), "NROScheduler %q failed to be deleted", nroSched.Name)
 }
 
 func WaitForMCPsCondition(cli client.Client, ctx context.Context, mcps []*machineconfigv1.MachineConfigPool, condition machineconfigv1.MachineConfigPoolConditionType) error {

test/utils/deploy/hypershift.go

@@ -39,7 +39,7 @@ const (
 	ConfigDataKey               = "config"
 )
 
-func (h *HyperShiftNRO) Deploy(ctx context.Context) *nropv1.NUMAResourcesOperator {
+func (h *HyperShiftNRO) Deploy(ctx context.Context, _ time.Duration) *nropv1.NUMAResourcesOperator {
 	GinkgoHelper()
 
 	hostedClusterName, err := hypershift.GetHostedClusterName()
@@ -105,7 +105,7 @@ func (h *HyperShiftNRO) Teardown(ctx context.Context, timeout time.Duration) {
 			Namespace: h.NroObj.Status.DaemonSets[0].Namespace,
 		}
 		Eventually(func() bool {
-			if err := e2eclient.Client.Get(context.TODO(), key, cm); !errors.IsNotFound(err) {
+			if err := e2eclient.Client.Get(ctx, key, cm); !errors.IsNotFound(err) {
 				if err == nil {
 					klog.Warningf("configmap %s still exists", key.String())
 				} else {
@@ -114,7 +114,7 @@ func (h *HyperShiftNRO) Teardown(ctx context.Context, timeout time.Duration) {
 				return false
 			}
 			return true
-		}).WithTimeout(5 * time.Minute).WithPolling(10 * time.Second).Should(BeTrue())
+		}).WithTimeout(timeout).WithPolling(10 * time.Second).Should(BeTrue())
 	}
 	Expect(e2eclient.Client.Delete(ctx, h.NroObj)).To(Succeed())
 }

test/utils/deploy/kubernetes.go

@@ -21,7 +21,7 @@ type KubernetesNRO struct {
 
 // Deploy returns a struct containing all the deployed objects,
 // so it will be easier to introspect and delete them later.
-func (k *KubernetesNRO) Deploy(ctx context.Context) *v1.NUMAResourcesOperator {
+func (k *KubernetesNRO) Deploy(ctx context.Context, timeout time.Duration) *v1.NUMAResourcesOperator {
 	GinkgoHelper()
 
 	mcpObj := objects.TestMCP()
@@ -31,7 +31,7 @@ func (k *KubernetesNRO) Deploy(ctx context.Context) *v1.NUMAResourcesOperator {
 	k.McpObj = mcpObj
 	matchLabels := map[string]string{"test": "test"}
 
-	return k.OpenShiftNRO.deployWithLabels(ctx, matchLabels)
+	return k.OpenShiftNRO.deployWithLabels(ctx, timeout, matchLabels)
 }
 
 func (k *KubernetesNRO) Teardown(ctx context.Context, timeout time.Duration) {

test/utils/deploy/openshift.go

@@ -35,13 +35,13 @@ type OpenShiftNRO struct {
 // Deploy deploys NUMAResourcesOperator object and
 // other dependencies, so the controller will be able to install TAS
 // stack properly
-func (o *OpenShiftNRO) Deploy(ctx context.Context) *nropv1.NUMAResourcesOperator {
+func (o *OpenShiftNRO) Deploy(ctx context.Context, timeout time.Duration) *nropv1.NUMAResourcesOperator {
 	GinkgoHelper()
 
-	return o.deployWithLabels(ctx, objects.OpenshiftMatchLabels())
+	return o.deployWithLabels(ctx, timeout, objects.OpenshiftMatchLabels())
 }
 
-func (o *OpenShiftNRO) deployWithLabels(ctx context.Context, matchLabels map[string]string) *nropv1.NUMAResourcesOperator {
+func (o *OpenShiftNRO) deployWithLabels(ctx context.Context, timeout time.Duration, matchLabels map[string]string) *nropv1.NUMAResourcesOperator {
 	GinkgoHelper()
 	nroObj := objects.TestNRO(objects.NROWithMCPSelector(matchLabels))
 	kcObj, err := objects.TestKC(matchLabels)
@@ -72,7 +72,7 @@ func (o *OpenShiftNRO) deployWithLabels(ctx context.Context, matchLabels map[str
 	o.NroObj = nroObj
 
 	By("unpausing the target MCPs")
-	Eventually(unpause).WithTimeout(configuration.MachineConfigPoolUpdateTimeout).WithPolling(configuration.MachineConfigPoolUpdateInterval).Should(Succeed())
+	Eventually(unpause).WithTimeout(timeout).WithPolling(configuration.MachineConfigPoolUpdateInterval).Should(Succeed())
 
 	By("updating the target NRO object")
 	Expect(e2eclient.Client.Get(ctx, client.ObjectKeyFromObject(nroObj), nroObj)).To(Succeed())