Merge pull request #8371 from blackpiglet/migration_case_support_vks

Make the E2E supporting VKS data mover environment.

test/Makefile

@@ -41,25 +41,41 @@ help:  ## Display this help
 
 TOOLS_DIR := $(REPO_ROOT)/hack/tools
 BIN_DIR := bin
+
 # Try to not modify PATH if possible
 GOBIN := $(REPO_ROOT)/.go/bin
+
 TOOLS_BIN_DIR := $(TOOLS_DIR)/$(BIN_DIR)
+
 GINKGO := $(GOBIN)/ginkgo
+
 KUSTOMIZE := $(TOOLS_BIN_DIR)/kustomize
+
 OUTPUT_DIR := _output/$(GOOS)/$(GOARCH)/bin
+
 # Please reference to this document for Ginkgo label spec format.
 # https://onsi.github.io/ginkgo/#spec-labels
 GINKGO_LABELS ?=
+
 # When --fail-fast is set, the entire suite will stop when the first failure occurs.
 # Enable --fail-fast by default.
 # https://onsi.github.io/ginkgo/#mental-model-how-ginkgo-handles-failure
 FAIL_FAST ?= false
+
 VELERO_CLI ?=$$(pwd)/../_output/bin/$(GOOS)/$(GOARCH)/velero
+
 VELERO_IMAGE ?= velero/velero:main
+
 PLUGINS ?=
+
+# Flag used to tell E2E whether the Velero vSphere plugin is installed.
+HAS_VSPHERE_PLUGIN ?= false
+
 RESTORE_HELPER_IMAGE ?=
+
 #Released version only
 UPGRADE_FROM_VELERO_VERSION ?= v1.13.2,v1.14.1
+
 # UPGRADE_FROM_VELERO_CLI can has the same format(a list divided by comma) with UPGRADE_FROM_VELERO_VERSION
 # Upgrade tests will be executed sequently according to the list by UPGRADE_FROM_VELERO_VERSION
 # So although length of UPGRADE_FROM_VELERO_CLI list is not equal with UPGRADE_FROM_VELERO_VERSION
@@ -150,7 +166,8 @@ COMMON_ARGS := --velerocli=$(VELERO_CLI) \
 	--velero-server-debug-mode=$(VELERO_SERVER_DEBUG_MODE) \
 	--uploader-type=$(UPLOADER_TYPE) \
 	--debug-velero-pod-restart=$(DEBUG_VELERO_POD_RESTART) \
-	--fail-fast=$(FAIL_FAST)
+	--fail-fast=$(FAIL_FAST) \
+	--has-vsphere-plugin=$(HAS_VSPHERE_PLUGIN)
 
 # Make sure ginkgo is in $GOBIN
 .PHONY:ginkgo

test/e2e/README.md

@@ -78,6 +78,7 @@ These configuration parameters are expected as values to the following command l
 1. `--standby-cluster-object-store-provider`: Object store provider for standby cluster.
 1. `--debug-velero-pod-restart`: A switch for debugging velero pod restart.
 1. `--fail-fast`: A switch for for failing fast on meeting error.
+1. `--has-vsphere-plugin`: A switch to indicate whether the Velero vSphere plugin is installed for vSphere environment.
 
 These configurations or parameters are used to generate install options for Velero for each test suite.
 
@@ -129,12 +130,13 @@ Below is a mapping between `make` variables to E2E configuration flags.
 1. `INSTALL_VELERO `: `-install-velero`. Optional.
 1. `DEBUG_VELERO_POD_RESTART`: `-debug-velero-pod-restart`. Optional.
 1. `FAIL_FAST`: `--fail-fast`. Optional.
+1. `HAS_VSPHERE_PLUGIN`: `--has-vsphere-plugin`. Optional.
 
 
 
 ### Examples
 
-Basic examples:
+#### Basic examples:
 
 1. Run Velero tests in a kind cluster with AWS (or MinIO) as the storage provider:
 
@@ -208,7 +210,7 @@ ADDITIONAL_CREDS_FILE=/path/to/azure-creds \
 make test-e2e
 ```
 
-Upgrade examples:
+#### Upgrade examples:
 
 1. Run Velero upgrade tests with pre-upgrade version:
 
@@ -234,7 +236,7 @@ UPGRADE_FROM_VELERO_VERSION=v1.10.2,v1.11.0 \
 make test-e2e
 ```
 
-Migration examples:
+#### Migration examples:
 
 1. Migration between 2 cluster of the same provider tests:
 
@@ -275,7 +277,7 @@ GINKGO_LABELS="Migration" \
 make test-e2e
 ```
 
-## 5. Filtering tests
+#### Filtering tests
 
 In release-1.15, Velero bumps the [Ginkgo](https://onsi.github.io/ginkgo/) version to [v2](https://onsi.github.io/ginkgo/MIGRATING_TO_V2).
 Velero E2E start to use [labels](https://onsi.github.io/ginkgo/#spec-labels) to filter cases instead of [`-focus` and `-skip`](https://onsi.github.io/ginkgo/#focused-specs) parameters.
@@ -285,7 +287,6 @@ Both `make run-e2e` and `make run-perf` CLI support using parameter `GINKGO_LABE
 `GINKGO_LABELS` is interpreted into `ginkgo run` CLI's parameter [`--label-filter`](https://onsi.github.io/ginkgo/#spec-labels).
 
 
-### Examples
 E2E tests can be run with specific cases to be included and/or excluded using the commands below:
 
 1. Run Velero tests with specific cases to be included:
@@ -316,6 +317,44 @@ In this example, cases are labelled as
 * `Migration` and `Restic` 
 will be skipped.
 
+#### VKS environment test
+1. Run the CSI data mover test. 
+
+`HAS_VSPHERE_PLUGIN` should be set to `false` to not install the Velero vSphere plugin.
+``` bash
+CLOUD_PROVIDER=vsphere \
+DEFAULT_CLUSTER=wl-antreav1301 \
+STANDBY_CLUSTER=wl-antreav1311 \
+DEFAULT_CLUSTER_NAME=192.168.0.4 \
+STANDBY_CLUSTER_NAME=192.168.0.3 \
+FEATURES=EnableCSI \
+PLUGINS=gcr.io/velero-gcp/velero-plugin-for-aws:main \
+HAS_VSPHERE_PLUGIN=false \
+OBJECT_STORE_PROVIDER=aws \
+CREDS_FILE=$HOME/aws-credential \
+BSL_CONFIG=region=us-east-1 \
+BSL_BUCKET=nightly-normal-account4-test \
+BSL_PREFIX=nightly \
+ADDITIONAL_BSL_PLUGINS=gcr.io/velero-gcp/velero-plugin-for-aws:main \
+ADDITIONAL_OBJECT_STORE_PROVIDER=aws \
+ADDITIONAL_BSL_CONFIG=region=us-east-1 \
+ADDITIONAL_BSL_BUCKET=nightly-restrict-account-test \
+ADDITIONAL_BSL_PREFIX=nightly \
+ADDITIONAL_CREDS_FILE=$HOME/aws-credential \
+VELERO_IMAGE=gcr.io/velero-gcp/velero:main \
+RESTORE_HELPER_IMAGE=gcr.io/velero-gcp/velero-restore-helper:main \
+VERSION=main \
+SNAPSHOT_MOVE_DATA=true \
+STANDBY_CLUSTER_CLOUD_PROVIDER=vsphere \
+STANDBY_CLUSTER_OBJECT_STORE_PROVIDER=aws \
+STANDBY_CLUSTER_PLUGINS=gcr.io/velero-gcp/velero-plugin-for-aws:main \
+DISABLE_INFORMER_CACHE=true \
+REGISTRY_CREDENTIAL_FILE=$HOME/.docker/config.json \
+GINKGO_LABELS=Migration \
+KIBISHII_DIRECTORY=$HOME/kibishii/kubernetes/yaml/ \
+make test-e2e
+```
+
 ## 6. Full Tests execution
 
 As we provided several examples for E2E test execution, if no filter is involved and despite difference of test environment, 

test/e2e/backup/backup.go

@@ -197,9 +197,9 @@ func BackupRestoreTest(backupRestoreTestConfig BackupRestoreTestConfig) {
 				secretKey,
 			)).To(Succeed())
 
-			bsls := []string{"default", additionalBsl}
+			BSLs := []string{"default", additionalBsl}
 
-			for _, bsl := range bsls {
+			for _, bsl := range BSLs {
 				backupName = fmt.Sprintf("backup-%s", bsl)
 				restoreName = fmt.Sprintf("restore-%s", bsl)
 				// We limit the length of backup name here to avoid the issue of vsphere plugin https://github.com/vmware-tanzu/velero-plugin-for-vsphere/issues/370
@@ -209,8 +209,8 @@ func BackupRestoreTest(backupRestoreTestConfig BackupRestoreTestConfig) {
 					restoreName = fmt.Sprintf("%s-%s", restoreName, UUIDgen)
 				}
 				veleroCfg.ProvideSnapshotsVolumeParam = !provideSnapshotVolumesParmInBackup
-				workloadNmespace := kibishiiNamespace + bsl
-				Expect(RunKibishiiTests(veleroCfg, backupName, restoreName, bsl, workloadNmespace, useVolumeSnapshots, !useVolumeSnapshots)).To(Succeed(),
+				workloadNS := kibishiiNamespace + bsl
+				Expect(RunKibishiiTests(veleroCfg, backupName, restoreName, bsl, workloadNS, useVolumeSnapshots, !useVolumeSnapshots)).To(Succeed(),
 					"Failed to successfully backup and restore Kibishii namespace using BSL %s", bsl)
 			}
 		})

test/e2e/backups/deletion.go

@@ -34,7 +34,7 @@ import (
 	. "github.com/vmware-tanzu/velero/test/util/velero"
 )
 
-// Test backup and restore of Kibishi using restic
+// Test backup and restore of Kibishii using restic
 
 func BackupDeletionWithSnapshots() {
 	backup_deletion_test(true)
@@ -99,8 +99,6 @@ func runBackupDeletionTests(client TestClient, veleroCfg VeleroConfig, backupLoc
 	providerName := veleroCfg.CloudProvider
 	veleroNamespace := veleroCfg.VeleroNamespace
 	registryCredentialFile := veleroCfg.RegistryCredentialFile
-	bslPrefix := veleroCfg.BSLPrefix
-	bslConfig := veleroCfg.BSLConfig
 	veleroFeatures := veleroCfg.Features
 	for _, ns := range workloadNamespaceList {
 		if err := CreateNamespace(oneHourTimeout, client, ns); err != nil {
@@ -143,7 +141,8 @@ func runBackupDeletionTests(client TestClient, veleroCfg VeleroConfig, backupLoc
 		})
 	})
 	for _, ns := range workloadNamespaceList {
-		if providerName == Vsphere && useVolumeSnapshots {
+		if useVolumeSnapshots &&
+			veleroCfg.HasVspherePlugin {
 			// Wait for uploads started by the Velero Plugin for vSphere to complete
 			// TODO - remove after upload progress monitoring is implemented
 			fmt.Println("Waiting for vSphere uploads to complete")
@@ -152,7 +151,7 @@ func runBackupDeletionTests(client TestClient, veleroCfg VeleroConfig, backupLoc
 			}
 		}
 	}
-	err = ObjectsShouldBeInBucket(veleroCfg.ObjectStoreProvider, veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, bslPrefix, bslConfig, backupName, BackupObjectsPrefix)
+	err = ObjectsShouldBeInBucket(veleroCfg.ObjectStoreProvider, veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, veleroCfg.BSLPrefix, veleroCfg.BSLConfig, backupName, BackupObjectsPrefix)
 	if err != nil {
 		return err
 	}
@@ -164,9 +163,12 @@ func runBackupDeletionTests(client TestClient, veleroCfg VeleroConfig, backupLoc
 			for _, ns := range workloadNamespaceList {
 				snapshotCheckPoint, err = GetSnapshotCheckPoint(client, veleroCfg, DefaultKibishiiWorkerCounts, ns, backupName, KibishiiPVCNameList)
 				Expect(err).NotTo(HaveOccurred(), "Fail to get Azure CSI snapshot checkpoint")
-				err = SnapshotsShouldBeCreatedInCloud(veleroCfg.CloudProvider,
-					veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, bslConfig,
-					backupName, snapshotCheckPoint)
+				err = CheckSnapshotsInProvider(
+					veleroCfg,
+					backupName,
+					snapshotCheckPoint,
+					false,
+				)
 				if err != nil {
 					return errors.Wrap(err, "exceed waiting for snapshot created in cloud")
 				}
@@ -178,9 +180,12 @@ func runBackupDeletionTests(client TestClient, veleroCfg VeleroConfig, backupLoc
 			Expect(err).NotTo(HaveOccurred(), "Fail to get Azure CSI snapshot checkpoint")
 
 			// Get all snapshots base on backup name, regardless of namespaces
-			err = SnapshotsShouldBeCreatedInCloud(veleroCfg.CloudProvider,
-				veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, bslConfig,
-				backupName, snapshotCheckPoint)
+			err = CheckSnapshotsInProvider(
+				veleroCfg,
+				backupName,
+				snapshotCheckPoint,
+				false,
+			)
 			if err != nil {
 				return errors.Wrap(err, "exceed waiting for snapshot created in cloud")
 			}
@@ -206,26 +211,34 @@ func runBackupDeletionTests(client TestClient, veleroCfg VeleroConfig, backupLoc
 		return err
 	}
 
+	// Verify snapshots are deleted after backup deletion.
 	if useVolumeSnapshots {
-		err = SnapshotsShouldNotExistInCloud(veleroCfg.CloudProvider,
-			veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, veleroCfg.BSLConfig,
-			backupName, snapshotCheckPoint)
+		snapshotCheckPoint.ExpectCount = 0
+		err = CheckSnapshotsInProvider(
+			veleroCfg,
+			backupName,
+			snapshotCheckPoint,
+			false,
+		)
 		if err != nil {
-			return errors.Wrap(err, "exceed waiting for snapshot created in cloud")
+			return errors.Wrap(err, "fail to verify snapshots are deleted in provider.")
 		}
 	}
 
-	err = ObjectsShouldNotBeInBucket(veleroCfg.ObjectStoreProvider, veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, bslPrefix, bslConfig, backupName, BackupObjectsPrefix, 5)
+	// Verify backup metadata files are deleted in OSS after backup deletion.
+	err = ObjectsShouldNotBeInBucket(
+		veleroCfg.ObjectStoreProvider,
+		veleroCfg.CloudCredentialsFile,
+		veleroCfg.BSLBucket,
+		veleroCfg.BSLPrefix,
+		veleroCfg.BSLConfig,
+		backupName,
+		BackupObjectsPrefix,
+		5,
+	)
 	if err != nil {
 		return err
 	}
-	if useVolumeSnapshots {
-		if err := SnapshotsShouldNotExistInCloud(veleroCfg.CloudProvider,
-			veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket,
-			bslConfig, backupName, snapshotCheckPoint); err != nil {
-			return errors.Wrap(err, "exceed waiting for snapshot created in cloud")
-		}
-	}
 
 	// Hit issue: https://docs.aws.amazon.com/AWSEC2/latest/APIReference/errors-overview.html#:~:text=SnapshotCreationPerVolumeRateExceeded
 	// Sleep for more than 15 seconds to avoid this issue.
@@ -242,13 +255,28 @@ func runBackupDeletionTests(client TestClient, veleroCfg VeleroConfig, backupLoc
 		})
 	})
 
-	err = DeleteObjectsInBucket(veleroCfg.ObjectStoreProvider, veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, bslPrefix, bslConfig, backupName, BackupObjectsPrefix)
-	if err != nil {
+	if err := DeleteObjectsInBucket(
+		veleroCfg.ObjectStoreProvider,
+		veleroCfg.CloudCredentialsFile,
+		veleroCfg.BSLBucket,
+		veleroCfg.BSLPrefix,
+		veleroCfg.BSLConfig,
+		backupName,
+		BackupObjectsPrefix,
+	); err != nil {
 		return err
 	}
 
-	err = ObjectsShouldNotBeInBucket(veleroCfg.ObjectStoreProvider, veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, bslPrefix, bslConfig, backupName, BackupObjectsPrefix, 1)
-	if err != nil {
+	if err := ObjectsShouldNotBeInBucket(
+		veleroCfg.ObjectStoreProvider,
+		veleroCfg.CloudCredentialsFile,
+		veleroCfg.BSLBucket,
+		veleroCfg.BSLPrefix,
+		veleroCfg.BSLConfig,
+		backupName,
+		BackupObjectsPrefix,
+		1,
+	); err != nil {
 		return err
 	}
 

test/e2e/backups/ttl.go

@@ -122,19 +122,31 @@ func TTLTest() {
 
 		var snapshotCheckPoint SnapshotCheckPoint
 		if useVolumeSnapshots {
-			if veleroCfg.CloudProvider == Vsphere {
-				// TODO - remove after upload progress monitoring is implemented
+			if veleroCfg.HasVspherePlugin {
 				By("Waiting for vSphere uploads to complete", func() {
 					Expect(WaitForVSphereUploadCompletion(ctx, time.Hour,
 						test.testNS, 2)).To(Succeed())
 				})
 			}
-			snapshotCheckPoint, err = GetSnapshotCheckPoint(client, veleroCfg, 2, test.testNS, test.backupName, KibishiiPVCNameList)
-			Expect(err).NotTo(HaveOccurred(), "Fail to get Azure CSI snapshot checkpoint")
 
-			Expect(SnapshotsShouldBeCreatedInCloud(veleroCfg.CloudProvider,
-				veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, veleroCfg.BSLConfig,
-				test.backupName, snapshotCheckPoint)).NotTo(HaveOccurred(), "Fail to get Azure CSI snapshot checkpoint")
+			snapshotCheckPoint, err = GetSnapshotCheckPoint(
+				client,
+				veleroCfg,
+				2,
+				test.testNS,
+				test.backupName,
+				KibishiiPVCNameList,
+			)
+			Expect(err).NotTo(HaveOccurred(), "Fail to get snapshot checkpoint")
+
+			Expect(
+				CheckSnapshotsInProvider(
+					veleroCfg,
+					test.backupName,
+					snapshotCheckPoint,
+					false,
+				),
+			).NotTo(HaveOccurred(), "Fail to verify the created snapshots")
 		}
 
 		By(fmt.Sprintf("Simulating a disaster by removing namespace %s\n", BackupCfg.BackupName), func() {
@@ -188,9 +200,13 @@ func TTLTest() {
 
 		By("PersistentVolume snapshots should be deleted", func() {
 			if useVolumeSnapshots {
-				Expect(SnapshotsShouldNotExistInCloud(veleroCfg.CloudProvider,
-					veleroCfg.CloudCredentialsFile, veleroCfg.BSLBucket, veleroCfg.BSLConfig,
-					test.backupName, snapshotCheckPoint)).NotTo(HaveOccurred(), "Fail to get Azure CSI snapshot checkpoint")
+				snapshotCheckPoint.ExpectCount = 0
+				Expect(CheckSnapshotsInProvider(
+					veleroCfg,
+					test.backupName,
+					snapshotCheckPoint,
+					false,
+				)).NotTo(HaveOccurred(), "Fail to get Azure CSI snapshot checkpoint")
 			}
 		})