Merge branch 'main' into backup-pvc-to-different-node

.github/workflows/e2e-test-kind.yaml

@@ -8,16 +8,26 @@ on:
       - "design/**"
       - "**/*.md"
 jobs:
+  get-go-version:
+    uses: ./.github/workflows/get-go-version.yaml
+    with:
+      ref: ${{ github.event.pull_request.base.ref }}
+
   # Build the Velero CLI and image once for all Kubernetes versions, and cache it so the fan-out workers can get it.
   build:
     runs-on: ubuntu-latest
+    needs: get-go-version
+    outputs:
+      minio-dockerfile-sha: ${{ steps.minio-version.outputs.dockerfile_sha }}
     steps:
       - name: Check out the code
         uses: actions/checkout@v5
-      - name: Set up Go
-        uses: actions/setup-go@v5
+      
+      - name: Set up Go version
+        uses: actions/setup-go@v6
         with:
-          go-version-file: 'go.mod'
+          go-version: ${{ needs.get-go-version.outputs.version }}
+
       # Look for a CLI that's made for this PR
       - name: Fetch built CLI
         id: cli-cache
@@ -44,6 +54,26 @@ jobs:
         run: |
           IMAGE=velero VERSION=pr-test BUILD_OUTPUT_TYPE=docker make container
           docker save velero:pr-test-linux-amd64 -o ./velero.tar
+      # Check and build MinIO image once for all e2e tests
+      - name: Check Bitnami MinIO Dockerfile version
+        id: minio-version
+        run: |
+          DOCKERFILE_SHA=$(curl -s https://api.github.com/repos/bitnami/containers/commits?path=bitnami/minio/2025/debian-12/Dockerfile\&per_page=1 | jq -r '.[0].sha')
+          echo "dockerfile_sha=${DOCKERFILE_SHA}" >> $GITHUB_OUTPUT
+      - name: Cache MinIO Image
+        uses: actions/cache@v4
+        id: minio-cache
+        with:
+          path: ./minio-image.tar
+          key: minio-bitnami-${{ steps.minio-version.outputs.dockerfile_sha }}
+      - name: Build MinIO Image from Bitnami Dockerfile
+        if: steps.minio-cache.outputs.cache-hit != 'true'
+        run: |
+          echo "Building MinIO image from Bitnami Dockerfile..."
+          git clone --depth 1 https://github.com/bitnami/containers.git /tmp/bitnami-containers
+          cd /tmp/bitnami-containers/bitnami/minio/2025/debian-12
+          docker build -t bitnami/minio:local .
+          docker save bitnami/minio:local > ${{ github.workspace }}/minio-image.tar
   # Create json of k8s versions to test
   # from guide: https://stackoverflow.com/a/65094398/4590470
   setup-test-matrix:
@@ -75,6 +105,7 @@ jobs:
     needs:
       - build
       - setup-test-matrix
+      - get-go-version
     runs-on: ubuntu-latest
     strategy:
       matrix: ${{fromJson(needs.setup-test-matrix.outputs.matrix)}}
@@ -82,13 +113,26 @@ jobs:
     steps:
       - name: Check out the code
         uses: actions/checkout@v5
-      - name: Set up Go
-        uses: actions/setup-go@v5
+
+      - name: Set up Go version
+        uses: actions/setup-go@v6
         with:
-          go-version-file: 'go.mod'
+          go-version: ${{ needs.get-go-version.outputs.version }}
+
+      # Fetch the pre-built MinIO image from the build job
+      - name: Fetch built MinIO Image
+        uses: actions/cache@v4
+        id: minio-cache
+        with:
+          path: ./minio-image.tar
+          key: minio-bitnami-${{ needs.build.outputs.minio-dockerfile-sha }}
+      - name: Load MinIO Image
+        run: |
+          echo "Loading MinIO image..."
+          docker load < ./minio-image.tar
       - name: Install MinIO
-        run:
-          docker run -d --rm -p 9000:9000 -e "MINIO_ACCESS_KEY=minio" -e "MINIO_SECRET_KEY=minio123" -e "MINIO_DEFAULT_BUCKETS=bucket,additional-bucket" bitnami/minio:2021.6.17-debian-10-r7
+        run: |
+          docker run -d --rm -p 9000:9000 -e "MINIO_ROOT_USER=minio" -e "MINIO_ROOT_PASSWORD=minio123" -e "MINIO_DEFAULT_BUCKETS=bucket,additional-bucket" bitnami/minio:local
       - uses: engineerd/setup-kind@v0.6.2
         with:
           skipClusterLogsExport: true

.github/workflows/get-go-version.yaml

@@ -0,0 +1,33 @@
+on:
+  workflow_call:
+    inputs:
+      ref:
+        description: "The target branch's ref"
+        required: true
+        type: string
+    outputs:
+      version: 
+        description: "The expected Go version"
+        value: ${{ jobs.extract.outputs.version }}
+
+jobs:
+  extract:
+      runs-on: ubuntu-latest
+      outputs:
+        version: ${{ steps.pick-version.outputs.version }}
+      steps:
+        - name: Check out the code
+          uses: actions/checkout@v5
+
+        - id: pick-version
+          run: |
+            if [ "${{ inputs.ref }}" == "main" ]; then
+              version=$(grep '^go ' go.mod | awk '{print $2}' | cut -d. -f1-2)
+            else
+              goDirectiveVersion=$(grep '^go ' go.mod | awk '{print $2}')
+              toolChainVersion=$(grep '^toolchain ' go.mod | awk '{print $2}')
+              version=$(printf "%s\n%s\n" "$goDirectiveVersion" "$toolChainVersion" | sort -V | tail -n1)
+            fi
+
+            echo "version=$version"
+            echo "version=$version" >> $GITHUB_OUTPUT

.github/workflows/pr-ci-check.yml

@@ -1,18 +1,26 @@
 name: Pull Request CI Check
 on: [pull_request]
 jobs:
+  get-go-version:
+    uses: ./.github/workflows/get-go-version.yaml
+    with:
+      ref: ${{ github.event.pull_request.base.ref }}
+
   build:
     name: Run CI
+    needs: get-go-version
     runs-on: ubuntu-latest
     strategy:
       fail-fast: false
     steps:
       - name: Check out the code
         uses: actions/checkout@v5
-      - name: Set up Go
-        uses: actions/setup-go@v5
+
+      - name: Set up Go version
+        uses: actions/setup-go@v6
         with:
-          go-version-file: 'go.mod'
+          go-version: ${{ needs.get-go-version.outputs.version }}      
+
       - name: Make ci
         run: make ci
       - name: Upload test coverage

.github/workflows/pr-linter-check.yml

@@ -7,16 +7,24 @@ on:
       - "design/**"
       - "**/*.md"
 jobs:
+  get-go-version:
+    uses: ./.github/workflows/get-go-version.yaml
+    with:
+      ref: ${{ github.event.pull_request.base.ref }}
+
   build:
     name: Run Linter Check
     runs-on: ubuntu-latest
+    needs: get-go-version
     steps:
       - name: Check out the code
         uses: actions/checkout@v5
-      - name: Set up Go
-        uses: actions/setup-go@v5
+
+      - name: Set up Go version
+        uses: actions/setup-go@v6
         with:
-          go-version-file: 'go.mod'
+          go-version: ${{ needs.get-go-version.outputs.version }}
+
       - name: Linter check
         uses: golangci/golangci-lint-action@v8
         with:

.github/workflows/push.yml

@@ -9,17 +9,24 @@ on:
       - '*'
 
 jobs:
+  get-go-version:
+    uses: ./.github/workflows/get-go-version.yaml
+    with:
+      ref: ${ github.ref }
 
   build:
     name: Build
     runs-on: ubuntu-latest
+    needs: get-go-version
     steps:
       - name: Check out the code
         uses: actions/checkout@v5
-      - name: Set up Go
-        uses: actions/setup-go@v5
+
+      - name: Set up Go version
+        uses: actions/setup-go@v6
         with:
-          go-version-file: 'go.mod'
+          go-version: ${{ needs.get-go-version.outputs.version }}
+
       - name: Set up QEMU
         id: qemu
         uses: docker/setup-qemu-action@v3

changelogs/unreleased/9248-0xLeo258

@@ -0,0 +1 @@
+Protect VolumeSnapshot field from race condition during multi-thread backup

changelogs/unreleased/9256-shubham-pampattiwar

@@ -0,0 +1 @@
+Fix repository maintenance jobs to inherit allowlisted tolerations from Velero deployment

changelogs/unreleased/9264-shubham-pampattiwar

@@ -0,0 +1 @@
+Fix schedule controller to prevent backup queue accumulation during extended blocking scenarios by properly handling empty backup phases

changelogs/unreleased/9281-0xLeo258

@@ -0,0 +1 @@
+Implement concurrency control for cache of native VolumeSnapshotter plugin.

pkg/backup/backup.go

@@ -366,7 +366,7 @@ func (kb *kubernetesBackupper) BackupWithResolvers(
 		discoveryHelper:          kb.discoveryHelper,
 		podVolumeBackupper:       podVolumeBackupper,
 		podVolumeSnapshotTracker: podvolume.NewTracker(),
-		volumeSnapshotterGetter:  volumeSnapshotterGetter,
+		volumeSnapshotterCache:   NewVolumeSnapshotterCache(volumeSnapshotterGetter),
 		itemHookHandler: &hook.DefaultItemHookHandler{
 			PodCommandExecutor: kb.podCommandExecutor,
 		},

pkg/backup/backup_test.go

@@ -3269,7 +3269,7 @@ func TestBackupWithSnapshots(t *testing.T) {
 			err := h.backupper.Backup(h.log, tc.req, backupFile, nil, nil, tc.snapshotterGetter)
 			require.NoError(t, err)
 
-			assert.Equal(t, tc.want, tc.req.VolumeSnapshots)
+			assert.Equal(t, tc.want, tc.req.VolumeSnapshots.Get())
 		})
 	}
 }
@@ -4213,7 +4213,7 @@ func TestBackupWithPodVolume(t *testing.T) {
 			assert.Equal(t, tc.want, req.PodVolumeBackups)
 
 			// this assumes that we don't have any test cases where some PVs should be snapshotted using a VolumeSnapshotter
-			assert.Nil(t, req.VolumeSnapshots)
+			assert.Nil(t, req.VolumeSnapshots.Get())
 		})
 	}
 }

pkg/backup/item_backupper.go

@@ -70,13 +70,11 @@ type itemBackupper struct {
 	discoveryHelper          discovery.Helper
 	podVolumeBackupper       podvolume.Backupper
 	podVolumeSnapshotTracker *podvolume.Tracker
-	volumeSnapshotterGetter  VolumeSnapshotterGetter
 	kubernetesBackupper      *kubernetesBackupper
-
-	itemHookHandler                    hook.ItemHookHandler
-	snapshotLocationVolumeSnapshotters map[string]vsv1.VolumeSnapshotter
-	hookTracker                        *hook.HookTracker
-	volumeHelperImpl                   volumehelper.VolumeHelper
+	volumeSnapshotterCache   *VolumeSnapshotterCache
+	itemHookHandler          hook.ItemHookHandler
+	hookTracker              *hook.HookTracker
+	volumeHelperImpl         volumehelper.VolumeHelper
 }
 
 type FileForArchive struct {
@@ -502,30 +500,6 @@ func (ib *itemBackupper) executeActions(
 	return obj, itemFiles, nil
 }
 
-// volumeSnapshotter instantiates and initializes a VolumeSnapshotter given a VolumeSnapshotLocation,
-// or returns an existing one if one's already been initialized for the location.
-func (ib *itemBackupper) volumeSnapshotter(snapshotLocation *velerov1api.VolumeSnapshotLocation) (vsv1.VolumeSnapshotter, error) {
-	if bs, ok := ib.snapshotLocationVolumeSnapshotters[snapshotLocation.Name]; ok {
-		return bs, nil
-	}
-
-	bs, err := ib.volumeSnapshotterGetter.GetVolumeSnapshotter(snapshotLocation.Spec.Provider)
-	if err != nil {
-		return nil, err
-	}
-
-	if err := bs.Init(snapshotLocation.Spec.Config); err != nil {
-		return nil, err
-	}
-
-	if ib.snapshotLocationVolumeSnapshotters == nil {
-		ib.snapshotLocationVolumeSnapshotters = make(map[string]vsv1.VolumeSnapshotter)
-	}
-	ib.snapshotLocationVolumeSnapshotters[snapshotLocation.Name] = bs
-
-	return bs, nil
-}
-
 // zoneLabelDeprecated is the label that stores availability-zone info
 // on PVs this is deprecated on Kubernetes >= 1.17.0
 // zoneLabel is the label that stores availability-zone info
@@ -641,7 +615,7 @@ func (ib *itemBackupper) takePVSnapshot(obj runtime.Unstructured, log logrus.Fie
 	for _, snapshotLocation := range ib.backupRequest.SnapshotLocations {
 		log := log.WithField("volumeSnapshotLocation", snapshotLocation.Name)
 
-		bs, err := ib.volumeSnapshotter(snapshotLocation)
+		bs, err := ib.volumeSnapshotterCache.SetNX(snapshotLocation)
 		if err != nil {
 			log.WithError(err).Error("Error getting volume snapshotter for volume snapshot location")
 			continue
@@ -699,7 +673,7 @@ func (ib *itemBackupper) takePVSnapshot(obj runtime.Unstructured, log logrus.Fie
 		snapshot.Status.Phase = volume.SnapshotPhaseCompleted
 		snapshot.Status.ProviderSnapshotID = snapshotID
 	}
-	ib.backupRequest.VolumeSnapshots = append(ib.backupRequest.VolumeSnapshots, snapshot)
+	ib.backupRequest.VolumeSnapshots.Add(snapshot)
 
 	// nil errors are automatically removed
 	return kubeerrs.NewAggregate(errs)

pkg/backup/request.go

@@ -17,6 +17,8 @@ limitations under the License.
 package backup
 
 import (
+	"sync"
+
 	"github.com/vmware-tanzu/velero/internal/hook"
 	"github.com/vmware-tanzu/velero/internal/resourcepolicies"
 	"github.com/vmware-tanzu/velero/internal/volume"
@@ -32,11 +34,27 @@ type itemKey struct {
 	name      string
 }
 
+type SynchronizedVSList struct {
+	sync.Mutex
+	VolumeSnapshotList []*volume.Snapshot
+}
+
+func (s *SynchronizedVSList) Add(vs *volume.Snapshot) {
+	s.Lock()
+	defer s.Unlock()
+	s.VolumeSnapshotList = append(s.VolumeSnapshotList, vs)
+}
+
+func (s *SynchronizedVSList) Get() []*volume.Snapshot {
+	s.Lock()
+	defer s.Unlock()
+	return s.VolumeSnapshotList
+}
+
 // Request is a request for a backup, with all references to other objects
 // materialized (e.g. backup/snapshot locations, includes/excludes, etc.)
 type Request struct {
 	*velerov1api.Backup
-
 	StorageLocation           *velerov1api.BackupStorageLocation
 	SnapshotLocations         []*velerov1api.VolumeSnapshotLocation
 	NamespaceIncludesExcludes *collections.IncludesExcludes
@@ -44,7 +62,7 @@ type Request struct {
 	ResourceHooks             []hook.ResourceHook
 	ResolvedActions           []framework.BackupItemResolvedActionV2
 	ResolvedItemBlockActions  []framework.ItemBlockResolvedAction
-	VolumeSnapshots           []*volume.Snapshot
+	VolumeSnapshots           SynchronizedVSList
 	PodVolumeBackups          []*velerov1api.PodVolumeBackup
 	BackedUpItems             *backedUpItemsMap
 	itemOperationsList        *[]*itemoperation.BackupOperation
@@ -80,7 +98,7 @@ func (r *Request) FillVolumesInformation() {
 	}
 
 	r.VolumesInformation.SkippedPVs = skippedPVMap
-	r.VolumesInformation.NativeSnapshots = r.VolumeSnapshots
+	r.VolumesInformation.NativeSnapshots = r.VolumeSnapshots.Get()
 	r.VolumesInformation.PodVolumeBackups = r.PodVolumeBackups
 	r.VolumesInformation.BackupOperations = *r.GetItemOperationsList()
 	r.VolumesInformation.BackupName = r.Backup.Name

pkg/backup/volume_snapshotter_cache.go

@@ -0,0 +1,42 @@
+package backup
+
+import (
+	"sync"
+
+	velerov1api "github.com/vmware-tanzu/velero/pkg/apis/velero/v1"
+	vsv1 "github.com/vmware-tanzu/velero/pkg/plugin/velero/volumesnapshotter/v1"
+)
+
+type VolumeSnapshotterCache struct {
+	cache  map[string]vsv1.VolumeSnapshotter
+	mutex  sync.Mutex
+	getter VolumeSnapshotterGetter
+}
+
+func NewVolumeSnapshotterCache(getter VolumeSnapshotterGetter) *VolumeSnapshotterCache {
+	return &VolumeSnapshotterCache{
+		cache:  make(map[string]vsv1.VolumeSnapshotter),
+		getter: getter,
+	}
+}
+
+func (c *VolumeSnapshotterCache) SetNX(location *velerov1api.VolumeSnapshotLocation) (vsv1.VolumeSnapshotter, error) {
+	c.mutex.Lock()
+	defer c.mutex.Unlock()
+
+	if snapshotter, exists := c.cache[location.Name]; exists {
+		return snapshotter, nil
+	}
+
+	snapshotter, err := c.getter.GetVolumeSnapshotter(location.Spec.Provider)
+	if err != nil {
+		return nil, err
+	}
+
+	if err := snapshotter.Init(location.Spec.Config); err != nil {
+		return nil, err
+	}
+
+	c.cache[location.Name] = snapshotter
+	return snapshotter, nil
+}

pkg/controller/backup_controller.go

@@ -734,8 +734,8 @@ func (b *backupReconciler) runBackup(backup *pkgbackup.Request) error {
 
 	// native snapshots phase will either be failed or completed right away
 	// https://github.com/vmware-tanzu/velero/blob/de3ea52f0cc478e99efa7b9524c7f353514261a4/pkg/backup/item_backupper.go#L632-L639
-	backup.Status.VolumeSnapshotsAttempted = len(backup.VolumeSnapshots)
-	for _, snap := range backup.VolumeSnapshots {
+	backup.Status.VolumeSnapshotsAttempted = len(backup.VolumeSnapshots.Get())
+	for _, snap := range backup.VolumeSnapshots.Get() {
 		if snap.Status.Phase == volume.SnapshotPhaseCompleted {
 			backup.Status.VolumeSnapshotsCompleted++
 		}
@@ -882,7 +882,7 @@ func persistBackup(backup *pkgbackup.Request,
 	}
 
 	// Velero-native volume snapshots (as opposed to CSI ones)
-	nativeVolumeSnapshots, errs := encode.ToJSONGzip(backup.VolumeSnapshots, "native volumesnapshots list")
+	nativeVolumeSnapshots, errs := encode.ToJSONGzip(backup.VolumeSnapshots.Get(), "native volumesnapshots list")
 	if errs != nil {
 		persistErrs = append(persistErrs, errs...)
 	}

pkg/controller/schedule_controller.go

@@ -229,7 +229,7 @@ func (c *scheduleReconciler) checkIfBackupInNewOrProgress(schedule *velerov1.Sch
 	}
 
 	for _, backup := range backupList.Items {
-		if backup.Status.Phase == velerov1.BackupPhaseNew || backup.Status.Phase == velerov1.BackupPhaseInProgress {
+		if backup.Status.Phase == "" || backup.Status.Phase == velerov1.BackupPhaseNew || backup.Status.Phase == velerov1.BackupPhaseInProgress {
 			log.Debugf("%s/%s still has backups that are in InProgress or New...", schedule.Namespace, schedule.Name)
 			return true
 		}

pkg/controller/schedule_controller_test.go

@@ -149,6 +149,13 @@ func TestReconcileOfSchedule(t *testing.T) {
 			expectedPhase: string(velerov1.SchedulePhaseEnabled),
 			backup:        builder.ForBackup("ns", "name-20220905120000").ObjectMeta(builder.WithLabels(velerov1.ScheduleNameLabel, "name")).Phase(velerov1.BackupPhaseNew).Result(),
 		},
+		{
+			name:          "schedule already has backup with empty phase (not yet reconciled).",
+			schedule:      newScheduleBuilder(velerov1.SchedulePhaseEnabled).CronSchedule("@every 5m").LastBackupTime("2000-01-01 00:00:00").Result(),
+			fakeClockTime: "2017-01-01 12:00:00",
+			expectedPhase: string(velerov1.SchedulePhaseEnabled),
+			backup:        builder.ForBackup("ns", "name-20220905120000").ObjectMeta(builder.WithLabels(velerov1.ScheduleNameLabel, "name")).Phase("").Result(),
+		},
 	}
 
 	for _, test := range tests {
@@ -215,10 +222,10 @@ func TestReconcileOfSchedule(t *testing.T) {
 			backups := &velerov1.BackupList{}
 			require.NoError(t, client.List(ctx, backups))
 
-			// If backup associated with schedule's status is in New or InProgress,
+			// If backup associated with schedule's status is in New or InProgress or empty phase,
 			// new backup shouldn't be submitted.
 			if test.backup != nil &&
-				(test.backup.Status.Phase == velerov1.BackupPhaseNew || test.backup.Status.Phase == velerov1.BackupPhaseInProgress) {
+				(test.backup.Status.Phase == "" || test.backup.Status.Phase == velerov1.BackupPhaseNew || test.backup.Status.Phase == velerov1.BackupPhaseInProgress) {
 				assert.Len(t, backups.Items, 1)
 				require.NoError(t, client.Delete(ctx, test.backup))
 			}
@@ -479,4 +486,19 @@ func TestCheckIfBackupInNewOrProgress(t *testing.T) {
 	reconciler = NewScheduleReconciler("namespace", logger, client, metrics.NewServerMetrics(), false)
 	result = reconciler.checkIfBackupInNewOrProgress(testSchedule)
 	assert.True(t, result)
+
+	// Clean backup in InProgress phase.
+	err = client.Delete(ctx, inProgressBackup)
+	require.NoError(t, err, "fail to delete backup in InProgress phase in TestCheckIfBackupInNewOrProgress: %v", err)
+
+	// Create backup with empty phase (not yet reconciled).
+	emptyPhaseBackup := builder.ForBackup("ns", "backup-3").
+		ObjectMeta(builder.WithLabels(velerov1.ScheduleNameLabel, "name")).
+		Phase("").Result()
+	err = client.Create(ctx, emptyPhaseBackup)
+	require.NoError(t, err, "fail to create backup with empty phase in TestCheckIfBackupInNewOrProgress: %v", err)
+
+	reconciler = NewScheduleReconciler("namespace", logger, client, metrics.NewServerMetrics(), false)
+	result = reconciler.checkIfBackupInNewOrProgress(testSchedule)
+	assert.True(t, result)
 }

pkg/repository/maintenance/maintenance.go

@@ -449,6 +449,35 @@ func StartNewJob(
 	return maintenanceJob.Name, nil
 }
 
+// buildTolerationsForMaintenanceJob builds the tolerations for maintenance jobs.
+// It includes the required Windows toleration for backward compatibility and filters
+// tolerations from the Velero deployment to only include those with keys that are
+// in the ThirdPartyTolerations allowlist, following the same pattern as labels and annotations.
+func buildTolerationsForMaintenanceJob(deployment *appsv1api.Deployment) []corev1api.Toleration {
+	// Start with the Windows toleration for backward compatibility
+	windowsToleration := corev1api.Toleration{
+		Key:      "os",
+		Operator: "Equal",
+		Effect:   "NoSchedule",
+		Value:    "windows",
+	}
+	result := []corev1api.Toleration{windowsToleration}
+
+	// Filter tolerations from the Velero deployment to only include allowed ones
+	// Only tolerations that exist on the deployment AND have keys in the allowlist are inherited
+	deploymentTolerations := veleroutil.GetTolerationsFromVeleroServer(deployment)
+	for _, k := range util.ThirdPartyTolerations {
+		for _, toleration := range deploymentTolerations {
+			if toleration.Key == k {
+				result = append(result, toleration)
+				break // Only add the first matching toleration for each allowed key
+			}
+		}
+	}
+
+	return result
+}
+
 func getPriorityClassName(ctx context.Context, cli client.Client, config *velerotypes.JobConfigs, logger logrus.FieldLogger) string {
 	// Use the priority class name from the global job configuration if available
 	// Note: Priority class is only read from global config, not per-repository
@@ -593,15 +622,8 @@ func buildJob(
 					SecurityContext:    podSecurityContext,
 					Volumes:            volumes,
 					ServiceAccountName: serviceAccount,
-					Tolerations: []corev1api.Toleration{
-						{
-							Key:      "os",
-							Operator: "Equal",
-							Effect:   "NoSchedule",
-							Value:    "windows",
-						},
-					},
-					ImagePullSecrets: imagePullSecrets,
+					Tolerations:        buildTolerationsForMaintenanceJob(deployment),
+					ImagePullSecrets:   imagePullSecrets,
 				},
 			},
 		},

pkg/repository/maintenance/maintenance_test.go

@@ -1481,3 +1481,291 @@ func TestBuildJobWithPriorityClassName(t *testing.T) {
 		})
 	}
 }
+
+func TestBuildTolerationsForMaintenanceJob(t *testing.T) {
+	windowsToleration := corev1api.Toleration{
+		Key:      "os",
+		Operator: "Equal",
+		Effect:   "NoSchedule",
+		Value:    "windows",
+	}
+
+	testCases := []struct {
+		name                  string
+		deploymentTolerations []corev1api.Toleration
+		expectedTolerations   []corev1api.Toleration
+	}{
+		{
+			name:                  "no tolerations should only include Windows toleration",
+			deploymentTolerations: nil,
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+			},
+		},
+		{
+			name:                  "empty tolerations should only include Windows toleration",
+			deploymentTolerations: []corev1api.Toleration{},
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+			},
+		},
+		{
+			name: "non-allowed toleration should not be inherited",
+			deploymentTolerations: []corev1api.Toleration{
+				{
+					Key:      "vng-ondemand",
+					Operator: "Equal",
+					Effect:   "NoSchedule",
+					Value:    "amd64",
+				},
+			},
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+			},
+		},
+		{
+			name: "allowed toleration should be inherited",
+			deploymentTolerations: []corev1api.Toleration{
+				{
+					Key:      "kubernetes.azure.com/scalesetpriority",
+					Operator: "Equal",
+					Effect:   "NoSchedule",
+					Value:    "spot",
+				},
+			},
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+				{
+					Key:      "kubernetes.azure.com/scalesetpriority",
+					Operator: "Equal",
+					Effect:   "NoSchedule",
+					Value:    "spot",
+				},
+			},
+		},
+		{
+			name: "mixed allowed and non-allowed tolerations should only inherit allowed",
+			deploymentTolerations: []corev1api.Toleration{
+				{
+					Key:      "vng-ondemand", // not in allowlist
+					Operator: "Equal",
+					Effect:   "NoSchedule",
+					Value:    "amd64",
+				},
+				{
+					Key:      "CriticalAddonsOnly", // in allowlist
+					Operator: "Exists",
+					Effect:   "NoSchedule",
+				},
+				{
+					Key:      "custom-key", // not in allowlist
+					Operator: "Equal",
+					Effect:   "NoSchedule",
+					Value:    "custom-value",
+				},
+			},
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+				{
+					Key:      "CriticalAddonsOnly",
+					Operator: "Exists",
+					Effect:   "NoSchedule",
+				},
+			},
+		},
+		{
+			name: "multiple allowed tolerations should all be inherited",
+			deploymentTolerations: []corev1api.Toleration{
+				{
+					Key:      "kubernetes.azure.com/scalesetpriority",
+					Operator: "Equal",
+					Effect:   "NoSchedule",
+					Value:    "spot",
+				},
+				{
+					Key:      "CriticalAddonsOnly",
+					Operator: "Exists",
+					Effect:   "NoSchedule",
+				},
+			},
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+				{
+					Key:      "kubernetes.azure.com/scalesetpriority",
+					Operator: "Equal",
+					Effect:   "NoSchedule",
+					Value:    "spot",
+				},
+				{
+					Key:      "CriticalAddonsOnly",
+					Operator: "Exists",
+					Effect:   "NoSchedule",
+				},
+			},
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			// Create a deployment with the specified tolerations
+			deployment := &appsv1api.Deployment{
+				Spec: appsv1api.DeploymentSpec{
+					Template: corev1api.PodTemplateSpec{
+						Spec: corev1api.PodSpec{
+							Tolerations: tc.deploymentTolerations,
+						},
+					},
+				},
+			}
+
+			result := buildTolerationsForMaintenanceJob(deployment)
+			assert.Equal(t, tc.expectedTolerations, result)
+		})
+	}
+}
+
+func TestBuildJobWithTolerationsInheritance(t *testing.T) {
+	// Define allowed tolerations that would be set on Velero deployment
+	allowedTolerations := []corev1api.Toleration{
+		{
+			Key:      "kubernetes.azure.com/scalesetpriority",
+			Operator: "Equal",
+			Effect:   "NoSchedule",
+			Value:    "spot",
+		},
+		{
+			Key:      "CriticalAddonsOnly",
+			Operator: "Exists",
+			Effect:   "NoSchedule",
+		},
+	}
+
+	// Mixed tolerations (allowed and non-allowed)
+	mixedTolerations := []corev1api.Toleration{
+		{
+			Key:      "vng-ondemand", // not in allowlist
+			Operator: "Equal",
+			Effect:   "NoSchedule",
+			Value:    "amd64",
+		},
+		{
+			Key:      "CriticalAddonsOnly", // in allowlist
+			Operator: "Exists",
+			Effect:   "NoSchedule",
+		},
+	}
+
+	// Windows toleration that should always be present
+	windowsToleration := corev1api.Toleration{
+		Key:      "os",
+		Operator: "Equal",
+		Effect:   "NoSchedule",
+		Value:    "windows",
+	}
+
+	testCases := []struct {
+		name                  string
+		deploymentTolerations []corev1api.Toleration
+		expectedTolerations   []corev1api.Toleration
+	}{
+		{
+			name:                  "no tolerations on deployment should only have Windows toleration",
+			deploymentTolerations: nil,
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+			},
+		},
+		{
+			name:                  "allowed tolerations should be inherited along with Windows toleration",
+			deploymentTolerations: allowedTolerations,
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+				{
+					Key:      "kubernetes.azure.com/scalesetpriority",
+					Operator: "Equal",
+					Effect:   "NoSchedule",
+					Value:    "spot",
+				},
+				{
+					Key:      "CriticalAddonsOnly",
+					Operator: "Exists",
+					Effect:   "NoSchedule",
+				},
+			},
+		},
+		{
+			name:                  "mixed tolerations should only inherit allowed ones",
+			deploymentTolerations: mixedTolerations,
+			expectedTolerations: []corev1api.Toleration{
+				windowsToleration,
+				{
+					Key:      "CriticalAddonsOnly",
+					Operator: "Exists",
+					Effect:   "NoSchedule",
+				},
+			},
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			// Create a new scheme and add necessary API types
+			localScheme := runtime.NewScheme()
+			err := velerov1api.AddToScheme(localScheme)
+			require.NoError(t, err)
+			err = appsv1api.AddToScheme(localScheme)
+			require.NoError(t, err)
+			err = batchv1api.AddToScheme(localScheme)
+			require.NoError(t, err)
+
+			// Create a deployment with the specified tolerations
+			deployment := &appsv1api.Deployment{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "velero",
+					Namespace: "velero",
+				},
+				Spec: appsv1api.DeploymentSpec{
+					Template: corev1api.PodTemplateSpec{
+						Spec: corev1api.PodSpec{
+							Containers: []corev1api.Container{
+								{
+									Name:  "velero",
+									Image: "velero/velero:latest",
+								},
+							},
+							Tolerations: tc.deploymentTolerations,
+						},
+					},
+				},
+			}
+
+			// Create a backup repository
+			repo := &velerov1api.BackupRepository{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "test-repo",
+					Namespace: "velero",
+				},
+				Spec: velerov1api.BackupRepositorySpec{
+					VolumeNamespace:       "velero",
+					BackupStorageLocation: "default",
+				},
+			}
+
+			// Create fake client and add the deployment
+			client := fake.NewClientBuilder().WithScheme(localScheme).WithObjects(deployment).Build()
+
+			// Create minimal job configs and resources
+			jobConfig := &velerotypes.JobConfigs{}
+			logLevel := logrus.InfoLevel
+			logFormat := logging.NewFormatFlag()
+			logFormat.Set("text")
+
+			// Call buildJob
+			job, err := buildJob(client, t.Context(), repo, "default", jobConfig, logLevel, logFormat, logrus.New())
+			require.NoError(t, err)
+
+			// Verify the tolerations are set correctly
+			assert.Equal(t, tc.expectedTolerations, job.Spec.Template.Spec.Tolerations)
+		})
+	}
+}