-Introduce volume group snapshot functionality include initializing queues and caches in snapshotter

- Introduce new flag for volume group snapshots and run worker if flag is enabled
- Introduce the main controller for group snapshots in snapshot-controller

cmd/snapshot-controller/main.go

@@ -72,6 +72,7 @@ var (
 	retryIntervalMax              = flag.Duration("retry-interval-max", 5*time.Minute, "Maximum retry interval of failed volume snapshot creation or deletion. Default is 5 minutes.")
 	enableDistributedSnapshotting = flag.Bool("enable-distributed-snapshotting", false, "Enables each node to handle snapshotting for the local volumes created on that node")
 	preventVolumeModeConversion   = flag.Bool("prevent-volume-mode-conversion", false, "Prevents an unauthorised user from modifying the volume mode when creating a PVC from an existing VolumeSnapshot.")
+	enableVolumeGroupSnapshots    = flag.Bool("enable-volume-group-snapshots", false, "Enables the volume group snapshot feature, allowing the user to create snapshots of groups of volumes.")
 
 	retryCRDIntervalMax = flag.Duration("retry-crd-interval-max", 5*time.Second, "Maximum retry interval to wait for CRDs to appear. The default is 5 seconds.")
 )
@@ -193,14 +194,20 @@ func main() {
 		factory.Snapshot().V1().VolumeSnapshots(),
 		factory.Snapshot().V1().VolumeSnapshotContents(),
 		factory.Snapshot().V1().VolumeSnapshotClasses(),
+		factory.Groupsnapshot().V1alpha1().VolumeGroupSnapshots(),
+		factory.Groupsnapshot().V1alpha1().VolumeGroupSnapshotContents(),
+		factory.Groupsnapshot().V1alpha1().VolumeGroupSnapshotClasses(),
 		coreFactory.Core().V1().PersistentVolumeClaims(),
 		nodeInformer,
 		metricsManager,
 		*resyncPeriod,
 		workqueue.NewItemExponentialFailureRateLimiter(*retryIntervalStart, *retryIntervalMax),
 		workqueue.NewItemExponentialFailureRateLimiter(*retryIntervalStart, *retryIntervalMax),
+		workqueue.NewItemExponentialFailureRateLimiter(*retryIntervalStart, *retryIntervalMax),
+		workqueue.NewItemExponentialFailureRateLimiter(*retryIntervalStart, *retryIntervalMax),
 		*enableDistributedSnapshotting,
 		*preventVolumeModeConversion,
+		*enableVolumeGroupSnapshots,
 	)
 
 	if err := ensureCustomResourceDefinitionsExist(snapClient); err != nil {

pkg/common-controller/framework_test.go

@@ -842,12 +842,18 @@ func newTestController(kubeClient kubernetes.Interface, clientset clientset.Inte
 		informerFactory.Snapshot().V1().VolumeSnapshots(),
 		informerFactory.Snapshot().V1().VolumeSnapshotContents(),
 		informerFactory.Snapshot().V1().VolumeSnapshotClasses(),
+		informerFactory.Groupsnapshot().V1alpha1().VolumeGroupSnapshots(),
+		informerFactory.Groupsnapshot().V1alpha1().VolumeGroupSnapshotContents(),
+		informerFactory.Groupsnapshot().V1alpha1().VolumeGroupSnapshotClasses(),
 		coreFactory.Core().V1().PersistentVolumeClaims(),
 		nil,
 		metricsManager,
 		60*time.Second,
 		workqueue.NewItemExponentialFailureRateLimiter(1*time.Millisecond, 1*time.Minute),
 		workqueue.NewItemExponentialFailureRateLimiter(1*time.Millisecond, 1*time.Minute),
+		workqueue.NewItemExponentialFailureRateLimiter(1*time.Millisecond, 1*time.Minute),
+		workqueue.NewItemExponentialFailureRateLimiter(1*time.Millisecond, 1*time.Minute),
+		false,
 		false,
 		false,
 	)

pkg/common-controller/groupsnapshot_controller_helper.go

@@ -0,0 +1,356 @@
+/*
+Copyright 2023 The Kubernetes Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package common_controller
+
+import (
+	"context"
+	"fmt"
+	"time"
+
+	v1 "k8s.io/api/core/v1"
+	"k8s.io/apimachinery/pkg/api/errors"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/labels"
+	klog "k8s.io/klog/v2"
+
+	crdv1alpha1 "github.com/kubernetes-csi/external-snapshotter/client/v6/apis/volumegroupsnapshot/v1alpha1"
+	crdv1 "github.com/kubernetes-csi/external-snapshotter/client/v6/apis/volumesnapshot/v1"
+	"github.com/kubernetes-csi/external-snapshotter/v6/pkg/utils"
+)
+
+func (ctrl *csiSnapshotCommonController) storeGroupSnapshotUpdate(groupsnapshot interface{}) (bool, error) {
+	return utils.StoreObjectUpdate(ctrl.groupSnapshotStore, groupsnapshot, "groupsnapshot")
+}
+
+func (ctrl *csiSnapshotCommonController) storeGroupSnapshotContentUpdate(groupsnapshotcontent interface{}) (bool, error) {
+	return utils.StoreObjectUpdate(ctrl.groupSnapshotContentStore, groupsnapshotcontent, "groupsnapshotcontent")
+}
+
+// getGroupSnapshotClass is a helper function to get group snapshot class from the class name.
+func (ctrl *csiSnapshotCommonController) getGroupSnapshotClass(className string) (*crdv1alpha1.VolumeGroupSnapshotClass, error) {
+	klog.V(5).Infof("getGroupSnapshotClass: VolumeGroupSnapshotClassName [%s]", className)
+
+	class, err := ctrl.groupSnapshotClassLister.Get(className)
+	if err != nil {
+		klog.Errorf("failed to retrieve group snapshot class %s from the informer: %q", className, err)
+		return nil, err
+	}
+
+	return class, nil
+}
+
+// updateGroupSnapshotErrorStatusWithEvent saves new groupsnapshot.Status to API
+// server and emits given event on the group snapshot. It saves the status and
+// emits the event only when the status has actually changed from the version
+// saved in API server.
+//
+// Parameters:
+//
+//   - groupSnapshot - group snapshot to update
+//   - setReadyToFalse bool - indicates whether to set the group snapshot's
+//     ReadyToUse status to false.
+//     if true, ReadyToUse will be set to false;
+//     otherwise, ReadyToUse will not be changed.
+//   - eventtype, reason, message - event to send, see EventRecorder.Event()
+func (ctrl *csiSnapshotCommonController) updateGroupSnapshotErrorStatusWithEvent(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot, setReadyToFalse bool, eventtype, reason, message string) error {
+	klog.V(5).Infof("updateGroupSnapshotErrorStatusWithEvent[%s]", utils.GroupSnapshotKey(groupSnapshot))
+
+	if groupSnapshot.Status != nil && groupSnapshot.Status.Error != nil && *groupSnapshot.Status.Error.Message == message {
+		klog.V(4).Infof("updateGroupSnapshotErrorStatusWithEvent[%s]: the same error %v is already set", groupSnapshot.Name, groupSnapshot.Status.Error)
+		return nil
+	}
+	groupSnapshotClone := groupSnapshot.DeepCopy()
+	if groupSnapshotClone.Status == nil {
+		groupSnapshotClone.Status = &crdv1alpha1.VolumeGroupSnapshotStatus{}
+	}
+	statusError := &crdv1.VolumeSnapshotError{
+		Time: &metav1.Time{
+			Time: time.Now(),
+		},
+		Message: &message,
+	}
+	groupSnapshotClone.Status.Error = statusError
+	// Only update ReadyToUse in VolumeSnapshot's Status to false if setReadyToFalse is true.
+	if setReadyToFalse {
+		ready := false
+		groupSnapshotClone.Status.ReadyToUse = &ready
+	}
+	newSnapshot, err := ctrl.clientset.GroupsnapshotV1alpha1().VolumeGroupSnapshots(groupSnapshotClone.Namespace).UpdateStatus(context.TODO(), groupSnapshotClone, metav1.UpdateOptions{})
+
+	// Emit the event even if the status update fails so that user can see the error
+	ctrl.eventRecorder.Event(newSnapshot, eventtype, reason, message)
+
+	if err != nil {
+		klog.V(4).Infof("updating VolumeGroupSnapshot[%s] error status failed %v", utils.GroupSnapshotKey(groupSnapshot), err)
+		return err
+	}
+
+	_, err = ctrl.storeGroupSnapshotUpdate(newSnapshot)
+	if err != nil {
+		klog.V(4).Infof("updating VolumeGroupSnapshot[%s] error status: cannot update internal cache %v", utils.GroupSnapshotKey(groupSnapshot), err)
+		return err
+	}
+
+	return nil
+}
+
+// SetDefaultGroupSnapshotClass is a helper function to figure out the default
+// group snapshot class.
+// For pre-provisioned case, it's an no-op.
+// For dynamic provisioning, it gets the default GroupSnapshotClasses in the
+// system if there is any (could be multiple), and finds the one with the same
+// CSI Driver as a PV from which a group snapshot will be taken.
+func (ctrl *csiSnapshotCommonController) SetDefaultGroupSnapshotClass(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) (*crdv1alpha1.VolumeGroupSnapshotClass, *crdv1alpha1.VolumeGroupSnapshot, error) {
+	klog.V(5).Infof("SetDefaultGroupSnapshotClass for group snapshot [%s]", groupSnapshot.Name)
+
+	if groupSnapshot.Spec.Source.VolumeGroupSnapshotContentName != nil {
+		// don't return error for pre-provisioned group snapshots
+		klog.V(5).Infof("Don't need to find GroupSnapshotClass for pre-provisioned group snapshot [%s]", groupSnapshot.Name)
+		return nil, groupSnapshot, nil
+	}
+
+	// Find default group snapshot class if available
+	list, err := ctrl.groupSnapshotClassLister.List(labels.Everything())
+	if err != nil {
+		return nil, groupSnapshot, err
+	}
+
+	pvDriver, err := ctrl.pvDriverFromGroupSnapshot(groupSnapshot)
+	if err != nil {
+		klog.Errorf("failed to get pv csi driver from group snapshot %s/%s: %q", groupSnapshot.Namespace, groupSnapshot.Name, err)
+		return nil, groupSnapshot, err
+	}
+
+	defaultClasses := []*crdv1alpha1.VolumeGroupSnapshotClass{}
+	for _, class := range list {
+		if utils.IsDefaultAnnotation(class.ObjectMeta) && pvDriver == class.Driver {
+			defaultClasses = append(defaultClasses, class)
+			klog.V(5).Infof("get defaultGroupClass added: %s, driver: %s", class.Name, pvDriver)
+		}
+	}
+	if len(defaultClasses) == 0 {
+		return nil, groupSnapshot, fmt.Errorf("cannot find default group snapshot class")
+	}
+	if len(defaultClasses) > 1 {
+		klog.V(4).Infof("get DefaultClass %d defaults found", len(defaultClasses))
+		return nil, groupSnapshot, fmt.Errorf("%d default snapshot classes were found", len(defaultClasses))
+	}
+	klog.V(5).Infof("setDefaultSnapshotClass [%s]: default VolumeSnapshotClassName [%s]", groupSnapshot.Name, defaultClasses[0].Name)
+	groupSnapshotClone := groupSnapshot.DeepCopy()
+	groupSnapshotClone.Spec.VolumeGroupSnapshotClassName = &(defaultClasses[0].Name)
+	newGroupSnapshot, err := ctrl.clientset.GroupsnapshotV1alpha1().VolumeGroupSnapshots(groupSnapshotClone.Namespace).Update(context.TODO(), groupSnapshotClone, metav1.UpdateOptions{})
+	if err != nil {
+		klog.V(4).Infof("updating VolumeSnapshot[%s] default class failed %v", utils.GroupSnapshotKey(groupSnapshot), err)
+	}
+	_, updateErr := ctrl.storeGroupSnapshotUpdate(newGroupSnapshot)
+	if updateErr != nil {
+		// We will get an "snapshot update" event soon, this is not a big error
+		klog.V(4).Infof("setDefaultSnapshotClass [%s]: cannot update internal cache: %v", utils.GroupSnapshotKey(groupSnapshot), updateErr)
+	}
+
+	return defaultClasses[0], newGroupSnapshot, nil
+}
+
+// pvDriverFromGroupSnapshot is a helper function to get the CSI driver name from the targeted PersistentVolume.
+// It looks up the PVC from which the snapshot is specified to be created from, and looks for the PVC's corresponding
+// PV. Bi-directional binding will be verified between PVC and PV before the PV's CSI driver is returned.
+// For an non-CSI volume, it returns an error immediately as it's not supported.
+func (ctrl *csiSnapshotCommonController) pvDriverFromGroupSnapshot(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) (string, error) {
+	pvs, err := ctrl.getVolumesFromVolumeGroupSnapshot(groupSnapshot)
+	if err != nil {
+		return "", err
+	}
+	// Take any volume to get the driver
+	if pvs[0].Spec.PersistentVolumeSource.CSI == nil {
+		return "", fmt.Errorf("snapshotting non-CSI volumes is not supported, snapshot:%s/%s", groupSnapshot.Namespace, groupSnapshot.Name)
+	}
+	return pvs[0].Spec.PersistentVolumeSource.CSI.Driver, nil
+}
+
+// getVolumesFromVolumeGroupSnapshot returns the list of PersistentVolume from a VolumeGroupSnapshot.
+func (ctrl *csiSnapshotCommonController) getVolumesFromVolumeGroupSnapshot(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) ([]*v1.PersistentVolume, error) {
+	var pvReturnList []*v1.PersistentVolume
+	pvcs, err := ctrl.getClaimsFromVolumeGroupSnapshot(groupSnapshot)
+	if err != nil {
+		return nil, err
+	}
+
+	for _, pvc := range pvcs {
+		if pvc.Status.Phase != v1.ClaimBound {
+			return nil, fmt.Errorf("the PVC %s is not yet bound to a PV, will not attempt to take a group snapshot", pvc.Name)
+		}
+
+		pvName := pvc.Spec.VolumeName
+		pv, err := ctrl.client.CoreV1().PersistentVolumes().Get(context.TODO(), pvName, metav1.GetOptions{})
+		if err != nil {
+			return nil, fmt.Errorf("failed to retrieve PV %s from the API server: %q", pvName, err)
+		}
+
+		// Verify binding between PV/PVC is still valid
+		bound := ctrl.isVolumeBoundToClaim(pv, &pvc)
+		if bound == false {
+			klog.Warningf("binding between PV %s and PVC %s is broken", pvName, pvc.Name)
+			return nil, fmt.Errorf("claim in dataSource not bound or invalid")
+		}
+		pvReturnList = append(pvReturnList, pv)
+		klog.V(5).Infof("getVolumeFromVolumeGroupSnapshot: group snapshot [%s] PV name [%s]", groupSnapshot.Name, pvName)
+	}
+
+	return pvReturnList, nil
+}
+
+// getClaimsFromVolumeGroupSnapshot is a helper function to get a list of PVCs from VolumeGroupSnapshot.
+func (ctrl *csiSnapshotCommonController) getClaimsFromVolumeGroupSnapshot(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) ([]v1.PersistentVolumeClaim, error) {
+	labelSelector := groupSnapshot.Spec.Source.Selector
+
+	// Get PVC that has group snapshot label applied.
+	pvcList, err := ctrl.client.CoreV1().PersistentVolumeClaims(groupSnapshot.Namespace).List(context.TODO(), metav1.ListOptions{LabelSelector: labelSelector.String()})
+	if err != nil {
+		return nil, fmt.Errorf("failed to list PVCs with label selector %s: %q", labelSelector.String(), err)
+	}
+	if len(pvcList.Items) == 0 {
+		return nil, fmt.Errorf("label selector %s for group snapshot not applied to any PVC", labelSelector.String())
+	}
+	return pvcList.Items, nil
+}
+
+// updateGroupSnapshot runs in worker thread and handles "groupsnapshot added",
+// "groupsnapshot updated" and "periodic sync" events.
+func (ctrl *csiSnapshotCommonController) updateGroupSnapshot(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) error {
+	// Store the new group snapshot version in the cache and do not process it
+	// if this is an old version.
+	klog.V(5).Infof("updateGroupSnapshot %q", utils.GroupSnapshotKey(groupSnapshot))
+	newGroupSnapshot, err := ctrl.storeGroupSnapshotUpdate(groupSnapshot)
+	if err != nil {
+		klog.Errorf("%v", err)
+	}
+	if !newGroupSnapshot {
+		return nil
+	}
+
+	err = ctrl.syncGroupSnapshot(groupSnapshot)
+	if err != nil {
+		if errors.IsConflict(err) {
+			// Version conflict error happens quite often and the controller
+			// recovers from it easily.
+			klog.V(3).Infof("could not sync snapshot %q: %+v", utils.GroupSnapshotKey(groupSnapshot), err)
+		} else {
+			klog.Errorf("could not sync snapshot %q: %+v", utils.GroupSnapshotKey(groupSnapshot), err)
+		}
+		return err
+	}
+	return nil
+}
+
+// deleteGroupSnapshot runs in worker thread and handles "groupsnapshot deleted" event.
+func (ctrl *csiSnapshotCommonController) deleteGroupSnapshot(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) {
+	_ = ctrl.snapshotStore.Delete(groupSnapshot)
+	klog.V(4).Infof("snapshot %q deleted", utils.GroupSnapshotKey(groupSnapshot))
+
+	groupSnapshotContentName := ""
+	if groupSnapshot.Status != nil && groupSnapshot.Status.BoundVolumeGroupSnapshotContentName != nil {
+		groupSnapshotContentName = *groupSnapshot.Status.BoundVolumeGroupSnapshotContentName
+	}
+	if groupSnapshotContentName == "" {
+		klog.V(5).Infof("deleteGroupSnapshot[%q]: group snapshot content not bound", utils.GroupSnapshotKey(groupSnapshot))
+		return
+	}
+
+	// sync the content when its group snapshot is deleted.  Explicitly sync'ing
+	// the content here in response to group snapshot deletion prevents the content
+	// from waiting until the next sync period for its release.
+	klog.V(5).Infof("deleteGroupSnapshot[%q]: scheduling sync of group snapshot content %s", utils.GroupSnapshotKey(groupSnapshot), groupSnapshotContentName)
+	ctrl.groupSnapshotContentQueue.Add(groupSnapshotContentName)
+}
+
+// syncGroupSnapshot is the main controller method to decide what to do with a
+// group snapshot. It's invoked by appropriate cache.Controller callbacks when
+// a group snapshot is created, updated or periodically synced. We do not
+// differentiate between these events.
+// For easier readability, it is split into syncUnreadyGroupSnapshot and syncReadyGroupSnapshot
+func (ctrl *csiSnapshotCommonController) syncGroupSnapshot(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) error {
+	klog.V(5).Infof("synchronizing VolumeGroupSnapshot[%s]", utils.GroupSnapshotKey(groupSnapshot))
+
+	klog.V(5).Infof("syncGroupSnapshot [%s]: check if we should remove finalizer on group snapshot PVC source and remove it if we can", utils.GroupSnapshotKey(groupSnapshot))
+
+	/*
+		TODO:
+		- Check and remove finalizer if needed.
+		- Check and set invalid group snapshot label, if needed.
+		- Process if deletion timestamp is set.
+		- Check and add group snapshot finalizers.
+	*/
+
+	// Need to build or update groupSnapshot.Status in following cases:
+	// 1) groupSnapshot.Status is nil
+	// 2) groupSnapshot.Status.ReadyToUse is false
+	// 3) groupSnapshot.Status.BoundVolumeSnapshotContentName is not set
+	if !utils.IsGroupSnapshotReady(groupSnapshot) || !utils.IsBoundVolumeGroupSnapshotContentNameSet(groupSnapshot) {
+		//return ctrl.syncUnreadyGroupSnapshot(groupSnapshot)
+	}
+	return ctrl.syncReadyGroupSnapshot(groupSnapshot)
+}
+
+// syncReadyGroupSnapshot checks the group snapshot which has been bound to group
+// snapshot content successfully before.
+// If there is any problem with the binding (e.g., group snapshot points to a
+// non-existent group snapshot content), update the group snapshot status and emit event.
+func (ctrl *csiSnapshotCommonController) syncReadyGroupSnapshot(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) error {
+	if !utils.IsBoundVolumeGroupSnapshotContentNameSet(groupSnapshot) {
+		return fmt.Errorf("group snapshot %s is not bound to a group snapshot content", utils.GroupSnapshotKey(groupSnapshot))
+	}
+	content, err := ctrl.getGroupSnapshotContentFromStore(*groupSnapshot.Status.BoundVolumeGroupSnapshotContentName)
+	if err != nil {
+		return nil
+	}
+	if content == nil {
+		// this meant there is no matching group snapshot content in cache found
+		// update status of the group snapshot and return
+		return ctrl.updateGroupSnapshotErrorStatusWithEvent(groupSnapshot, true, v1.EventTypeWarning, "GroupSnapshotContentMissing", "VolumeGroupSnapshotContent is missing")
+	}
+	klog.V(5).Infof("syncReadyGroupSnapshot[%s]: VolumeGroupSnapshotContent %q found", utils.GroupSnapshotKey(groupSnapshot), content.Name)
+	// check binding from group snapshot content side to make sure the binding is still valid
+	if !utils.IsVolumeGroupSnapshotRefSet(groupSnapshot, content) {
+		// group snapshot is bound but group snapshot content is not pointing to the group snapshot
+		return ctrl.updateGroupSnapshotErrorStatusWithEvent(groupSnapshot, true, v1.EventTypeWarning, "GroupSnapshotMisbound", "VolumeGroupSnapshotContent is not bound to the VolumeGroupSnapshot correctly")
+	}
+
+	// everything is verified, return
+	return nil
+}
+
+// getGroupSnapshotContentFromStore tries to find a VolumeGroupSnapshotContent from group
+// snapshot content cache store by name.
+// Note that if no VolumeGroupSnapshotContent exists in the cache store and no error
+// encountered, it returns (nil, nil)
+func (ctrl *csiSnapshotCommonController) getGroupSnapshotContentFromStore(contentName string) (*crdv1alpha1.VolumeGroupSnapshotContent, error) {
+	obj, exist, err := ctrl.groupSnapshotContentStore.GetByKey(contentName)
+	if err != nil {
+		// should never reach here based on implementation at:
+		// https://github.com/kubernetes/client-go/blob/master/tools/cache/store.go#L226
+		return nil, err
+	}
+	if !exist {
+		// not able to find a matching content
+		return nil, nil
+	}
+	content, ok := obj.(*crdv1alpha1.VolumeGroupSnapshotContent)
+	if !ok {
+		return nil, fmt.Errorf("expected VolumeSnapshotContent, got %+v", obj)
+	}
+	return content, nil
+}

pkg/common-controller/snapshot_controller_base.go

@@ -20,10 +20,13 @@ import (
 	"fmt"
 	"time"
 
+	crdv1alpha1 "github.com/kubernetes-csi/external-snapshotter/client/v6/apis/volumegroupsnapshot/v1alpha1"
 	crdv1 "github.com/kubernetes-csi/external-snapshotter/client/v6/apis/volumesnapshot/v1"
 	clientset "github.com/kubernetes-csi/external-snapshotter/client/v6/clientset/versioned"
-	storageinformers "github.com/kubernetes-csi/external-snapshotter/client/v6/informers/externalversions/volumesnapshot/v1"
+	groupsnapshotinformers "github.com/kubernetes-csi/external-snapshotter/client/v6/informers/externalversions/volumegroupsnapshot/v1alpha1"
-	storagelisters "github.com/kubernetes-csi/external-snapshotter/client/v6/listers/volumesnapshot/v1"
+	snapshotinformers "github.com/kubernetes-csi/external-snapshotter/client/v6/informers/externalversions/volumesnapshot/v1"
+	groupsnapshotlisters "github.com/kubernetes-csi/external-snapshotter/client/v6/listers/volumegroupsnapshot/v1alpha1"
+	snapshotlisters "github.com/kubernetes-csi/external-snapshotter/client/v6/listers/volumesnapshot/v1"
 	"github.com/kubernetes-csi/external-snapshotter/v6/pkg/metrics"
 	"github.com/kubernetes-csi/external-snapshotter/v6/pkg/utils"
 
@@ -48,20 +51,30 @@ type csiSnapshotCommonController struct {
 	eventRecorder             record.EventRecorder
 	snapshotQueue             workqueue.RateLimitingInterface
 	contentQueue              workqueue.RateLimitingInterface
+	groupSnapshotQueue        workqueue.RateLimitingInterface
+	groupSnapshotContentQueue workqueue.RateLimitingInterface
 
-	snapshotLister       storagelisters.VolumeSnapshotLister
+	snapshotLister                   snapshotlisters.VolumeSnapshotLister
 	snapshotListerSynced             cache.InformerSynced
-	contentLister        storagelisters.VolumeSnapshotContentLister
+	contentLister                    snapshotlisters.VolumeSnapshotContentLister
 	contentListerSynced              cache.InformerSynced
-	classLister          storagelisters.VolumeSnapshotClassLister
+	classLister                      snapshotlisters.VolumeSnapshotClassLister
 	classListerSynced                cache.InformerSynced
 	pvcLister                        corelisters.PersistentVolumeClaimLister
 	pvcListerSynced                  cache.InformerSynced
 	nodeLister                       corelisters.NodeLister
 	nodeListerSynced                 cache.InformerSynced
+	groupSnapshotLister              groupsnapshotlisters.VolumeGroupSnapshotLister
+	groupSnapshotListerSynced        cache.InformerSynced
+	groupSnapshotContentLister       groupsnapshotlisters.VolumeGroupSnapshotContentLister
+	groupSnapshotContentListerSynced cache.InformerSynced
+	groupSnapshotClassLister         groupsnapshotlisters.VolumeGroupSnapshotClassLister
+	groupSnapshotClassListerSynced   cache.InformerSynced
 
 	snapshotStore             cache.Store
 	contentStore              cache.Store
+	groupSnapshotStore        cache.Store
+	groupSnapshotContentStore cache.Store
 
 	metricsManager metrics.MetricsManager
 
@@ -69,23 +82,30 @@ type csiSnapshotCommonController struct {
 
 	enableDistributedSnapshotting bool
 	preventVolumeModeConversion   bool
+	enableVolumeGroupSnapshots    bool
 }
 
 // NewCSISnapshotController returns a new *csiSnapshotCommonController
 func NewCSISnapshotCommonController(
 	clientset clientset.Interface,
 	client kubernetes.Interface,
-	volumeSnapshotInformer storageinformers.VolumeSnapshotInformer,
+	volumeSnapshotInformer snapshotinformers.VolumeSnapshotInformer,
-	volumeSnapshotContentInformer storageinformers.VolumeSnapshotContentInformer,
+	volumeSnapshotContentInformer snapshotinformers.VolumeSnapshotContentInformer,
-	volumeSnapshotClassInformer storageinformers.VolumeSnapshotClassInformer,
+	volumeSnapshotClassInformer snapshotinformers.VolumeSnapshotClassInformer,
+	volumeGroupSnapshotInformer groupsnapshotinformers.VolumeGroupSnapshotInformer,
+	volumeGroupSnapshotContentInformer groupsnapshotinformers.VolumeGroupSnapshotContentInformer,
+	volumeGroupSnapshotClassInformer groupsnapshotinformers.VolumeGroupSnapshotClassInformer,
 	pvcInformer coreinformers.PersistentVolumeClaimInformer,
 	nodeInformer coreinformers.NodeInformer,
 	metricsManager metrics.MetricsManager,
 	resyncPeriod time.Duration,
 	snapshotRateLimiter workqueue.RateLimiter,
 	contentRateLimiter workqueue.RateLimiter,
+	groupSnapshotRateLimiter workqueue.RateLimiter,
+	groupSnapshotContentRateLimiter workqueue.RateLimiter,
 	enableDistributedSnapshotting bool,
 	preventVolumeModeConversion bool,
+	enableVolumeGroupSnapshots bool,
 ) *csiSnapshotCommonController {
 	broadcaster := record.NewBroadcaster()
 	broadcaster.StartLogging(klog.Infof)
@@ -142,12 +162,49 @@ func NewCSISnapshotCommonController(
 
 	ctrl.preventVolumeModeConversion = preventVolumeModeConversion
 
+	ctrl.enableVolumeGroupSnapshots = enableVolumeGroupSnapshots
+
+	if enableVolumeGroupSnapshots {
+		ctrl.groupSnapshotQueue = workqueue.NewNamedRateLimitingQueue(groupSnapshotRateLimiter, "snapshot-controller-group-snapshot")
+		ctrl.groupSnapshotContentQueue = workqueue.NewNamedRateLimitingQueue(groupSnapshotContentRateLimiter, "snapshot-controller-group-content")
+
+		volumeGroupSnapshotInformer.Informer().AddEventHandlerWithResyncPeriod(
+			cache.ResourceEventHandlerFuncs{
+				AddFunc:    func(obj interface{}) { ctrl.enqueueGroupSnapshotWork(obj) },
+				UpdateFunc: func(oldObj, newObj interface{}) { ctrl.enqueueGroupSnapshotWork(newObj) },
+				DeleteFunc: func(obj interface{}) { ctrl.enqueueGroupSnapshotWork(obj) },
+			},
+			ctrl.resyncPeriod,
+		)
+		ctrl.groupSnapshotLister = volumeGroupSnapshotInformer.Lister()
+		ctrl.groupSnapshotListerSynced = volumeGroupSnapshotInformer.Informer().HasSynced
+
+		volumeGroupSnapshotContentInformer.Informer().AddEventHandlerWithResyncPeriod(
+			cache.ResourceEventHandlerFuncs{
+				AddFunc:    func(obj interface{}) { ctrl.enqueueGroupSnapshotContentWork(obj) },
+				UpdateFunc: func(oldObj, newObj interface{}) { ctrl.enqueueGroupSnapshotContentWork(newObj) },
+				DeleteFunc: func(obj interface{}) { ctrl.enqueueGroupSnapshotContentWork(obj) },
+			},
+			ctrl.resyncPeriod,
+		)
+		ctrl.groupSnapshotContentLister = volumeGroupSnapshotContentInformer.Lister()
+		ctrl.groupSnapshotContentListerSynced = volumeGroupSnapshotContentInformer.Informer().HasSynced
+
+		ctrl.groupSnapshotClassLister = volumeGroupSnapshotClassInformer.Lister()
+		ctrl.groupSnapshotClassListerSynced = volumeGroupSnapshotClassInformer.Informer().HasSynced
+
+	}
+
 	return ctrl
 }
 
 func (ctrl *csiSnapshotCommonController) Run(workers int, stopCh <-chan struct{}) {
 	defer ctrl.snapshotQueue.ShutDown()
 	defer ctrl.contentQueue.ShutDown()
+	if ctrl.enableVolumeGroupSnapshots {
+		defer ctrl.groupSnapshotQueue.ShutDown()
+		defer ctrl.groupSnapshotContentQueue.ShutDown()
+	}
 
 	klog.Infof("Starting snapshot controller")
 	defer klog.Infof("Shutting snapshot controller")
@@ -156,17 +213,24 @@ func (ctrl *csiSnapshotCommonController) Run(workers int, stopCh <-chan struct{}
 	if ctrl.enableDistributedSnapshotting {
 		informersSynced = append(informersSynced, ctrl.nodeListerSynced)
 	}
+	if ctrl.enableVolumeGroupSnapshots {
+		informersSynced = append(informersSynced, []cache.InformerSynced{ctrl.groupSnapshotListerSynced, ctrl.groupSnapshotContentListerSynced, ctrl.groupSnapshotClassListerSynced}...)
+	}
 
 	if !cache.WaitForCacheSync(stopCh, informersSynced...) {
 		klog.Errorf("Cannot sync caches")
 		return
 	}
 
-	ctrl.initializeCaches(ctrl.snapshotLister, ctrl.contentLister)
+	ctrl.initializeCaches()
 
 	for i := 0; i < workers; i++ {
 		go wait.Until(ctrl.snapshotWorker, 0, stopCh)
 		go wait.Until(ctrl.contentWorker, 0, stopCh)
+		if ctrl.enableVolumeGroupSnapshots {
+			go wait.Until(ctrl.groupSnapshotWorker, 0, stopCh)
+			go wait.Until(ctrl.groupSnapshotContentWorker, 0, stopCh)
+		}
 	}
 
 	<-stopCh
@@ -481,8 +545,8 @@ func (ctrl *csiSnapshotCommonController) deleteContent(content *crdv1.VolumeSnap
 // initializeCaches fills all controller caches with initial data from etcd in
 // order to have the caches already filled when first addSnapshot/addContent to
 // perform initial synchronization of the controller.
-func (ctrl *csiSnapshotCommonController) initializeCaches(snapshotLister storagelisters.VolumeSnapshotLister, contentLister storagelisters.VolumeSnapshotContentLister) {
+func (ctrl *csiSnapshotCommonController) initializeCaches() {
-	snapshotList, err := snapshotLister.List(labels.Everything())
+	snapshotList, err := ctrl.snapshotLister.List(labels.Everything())
 	if err != nil {
 		klog.Errorf("CSISnapshotController can't initialize caches: %v", err)
 		return
@@ -494,7 +558,7 @@ func (ctrl *csiSnapshotCommonController) initializeCaches(snapshotLister storage
 		}
 	}
 
-	contentList, err := contentLister.List(labels.Everything())
+	contentList, err := ctrl.contentLister.List(labels.Everything())
 	if err != nil {
 		klog.Errorf("CSISnapshotController can't initialize caches: %v", err)
 		return
@@ -506,5 +570,195 @@ func (ctrl *csiSnapshotCommonController) initializeCaches(snapshotLister storage
 		}
 	}
 
+	if ctrl.enableVolumeGroupSnapshots {
+		groupSnapshotList, err := ctrl.snapshotLister.List(labels.Everything())
+		if err != nil {
+			klog.Errorf("CSISnapshotController can't initialize caches: %v", err)
+			return
+		}
+		for _, groupSnapshot := range groupSnapshotList {
+			groupSnapshotClone := groupSnapshot.DeepCopy()
+			if _, err = ctrl.storeGroupSnapshotUpdate(groupSnapshotClone); err != nil {
+				klog.Errorf("error updating volume group snapshot cache: %v", err)
+			}
+		}
+
+		groupContentList, err := ctrl.contentLister.List(labels.Everything())
+		if err != nil {
+			klog.Errorf("CSISnapshotController can't initialize caches: %v", err)
+			return
+		}
+		for _, groupContent := range groupContentList {
+			groupContentClone := groupContent.DeepCopy()
+			if _, err = ctrl.storeGroupSnapshotContentUpdate(groupContentClone); err != nil {
+				klog.Errorf("error updating volume group snapshot content cache: %v", err)
+			}
+		}
+	}
+
 	klog.V(4).Infof("controller initialized")
 }
+
+// enqueueGroupSnapshotWork adds group snapshot to given work queue.
+func (ctrl *csiSnapshotCommonController) enqueueGroupSnapshotWork(obj interface{}) {
+	// Beware of "xxx deleted" events
+	if unknown, ok := obj.(cache.DeletedFinalStateUnknown); ok && unknown.Obj != nil {
+		obj = unknown.Obj
+	}
+	if groupSnapshot, ok := obj.(*crdv1alpha1.VolumeGroupSnapshot); ok {
+		objName, err := cache.DeletionHandlingMetaNamespaceKeyFunc(groupSnapshot)
+		if err != nil {
+			klog.Errorf("failed to get key from object: %v, %v", err, groupSnapshot)
+			return
+		}
+		klog.V(5).Infof("enqueued %q for sync", objName)
+		ctrl.groupSnapshotQueue.Add(objName)
+	}
+}
+
+// enqueueGroupSnapshotContentWork adds group snapshot content to given work queue.
+func (ctrl *csiSnapshotCommonController) enqueueGroupSnapshotContentWork(obj interface{}) {
+	// Beware of "xxx deleted" events
+	if unknown, ok := obj.(cache.DeletedFinalStateUnknown); ok && unknown.Obj != nil {
+		obj = unknown.Obj
+	}
+	if content, ok := obj.(*crdv1alpha1.VolumeGroupSnapshotContent); ok {
+		objName, err := cache.DeletionHandlingMetaNamespaceKeyFunc(content)
+		if err != nil {
+			klog.Errorf("failed to get key from object: %v, %v", err, content)
+			return
+		}
+		klog.V(5).Infof("enqueued %q for sync", objName)
+		ctrl.groupSnapshotContentQueue.Add(objName)
+	}
+}
+
+// groupSnapshotWorker is the main worker for VolumeGroupSnapshots.
+func (ctrl *csiSnapshotCommonController) groupSnapshotWorker() {
+	keyObj, quit := ctrl.groupSnapshotQueue.Get()
+	if quit {
+		return
+	}
+	defer ctrl.groupSnapshotQueue.Done(keyObj)
+
+	if err := ctrl.syncGroupSnapshotByKey(keyObj.(string)); err != nil {
+		// Rather than wait for a full resync, re-add the key to the
+		// queue to be processed.
+		ctrl.groupSnapshotQueue.AddRateLimited(keyObj)
+		klog.V(4).Infof("Failed to sync group snapshot %q, will retry again: %v", keyObj.(string), err)
+	} else {
+		// Finally, if no error occurs we forget this item so it does not
+		// get queued again until another change happens.
+		ctrl.groupSnapshotQueue.Forget(keyObj)
+	}
+}
+
+// groupSnapshotContentWorker is the main worker for VolumeGroupSnapshotContent.
+func (ctrl *csiSnapshotCommonController) groupSnapshotContentWorker() {
+	keyObj, quit := ctrl.groupSnapshotContentQueue.Get()
+	if quit {
+		return
+	}
+	defer ctrl.groupSnapshotContentQueue.Done(keyObj)
+
+	if err := ctrl.syncContentByKey(keyObj.(string)); err != nil {
+		// Rather than wait for a full resync, re-add the key to the
+		// queue to be processed.
+		ctrl.groupSnapshotContentQueue.AddRateLimited(keyObj)
+		klog.V(4).Infof("Failed to sync content %q, will retry again: %v", keyObj.(string), err)
+	} else {
+		// Finally, if no error occurs we Forget this item so it does not
+		// get queued again until another change happens.
+		ctrl.groupSnapshotContentQueue.Forget(keyObj)
+	}
+}
+
+// syncGroupSnapshotByKey processes a VolumeGroupSnapshot request.
+func (ctrl *csiSnapshotCommonController) syncGroupSnapshotByKey(key string) error {
+	klog.V(5).Infof("syncGroupSnapshotByKey[%s]", key)
+
+	namespace, name, err := cache.SplitMetaNamespaceKey(key)
+	klog.V(5).Infof("groupSnapshotWorker: group snapshot namespace [%s] name [%s]", namespace, name)
+	if err != nil {
+		klog.Errorf("error getting namespace & name of group snapshot %q to get group snapshot from informer: %v", key, err)
+		return nil
+	}
+	groupSnapshot, err := ctrl.groupSnapshotLister.VolumeGroupSnapshots(namespace).Get(name)
+	if err == nil {
+		// The volume group snapshot still exists in informer cache, the event must have
+		// been add/update/sync
+		newGroupSnapshot, err := ctrl.checkAndUpdateGroupSnapshotClass(groupSnapshot)
+		if err == nil || (newGroupSnapshot.ObjectMeta.DeletionTimestamp != nil && errors.IsNotFound(err)) {
+			// If the VolumeSnapshotClass is not found, we still need to process an update
+			// so that syncGroupSnapshot can delete the snapshot, should it still exist in the
+			// cluster after it's been removed from the informer cache
+			if newGroupSnapshot.ObjectMeta.DeletionTimestamp != nil && errors.IsNotFound(err) {
+				klog.V(5).Infof("GroupSnapshot %q is being deleted. GroupSnapshotClass has already been removed", key)
+			}
+			klog.V(5).Infof("Updating group snapshot %q", key)
+			return ctrl.updateGroupSnapshot(newGroupSnapshot)
+		}
+		return err
+	}
+	if err != nil && !errors.IsNotFound(err) {
+		klog.V(2).Infof("error getting group snapshot %q from informer: %v", key, err)
+		return err
+	}
+	// The group snapshot is not in informer cache, the event must have been "delete"
+	vgsObj, found, err := ctrl.groupSnapshotStore.GetByKey(key)
+	if err != nil {
+		klog.V(2).Infof("error getting group snapshot %q from cache: %v", key, err)
+		return nil
+	}
+	if !found {
+		// The controller has already processed the delete event and
+		// deleted the group snapshot from its cache
+		klog.V(2).Infof("deletion of group snapshot %q was already processed", key)
+		return nil
+	}
+	groupSnapshot, ok := vgsObj.(*crdv1alpha1.VolumeGroupSnapshot)
+	if !ok {
+		klog.Errorf("expected vgs, got %+v", vgsObj)
+		return nil
+	}
+
+	klog.V(5).Infof("deleting group snapshot %q", key)
+	ctrl.deleteGroupSnapshot(groupSnapshot)
+
+	return nil
+}
+
+// checkAndUpdateGroupSnapshotClass gets the VolumeGroupSnapshotClass from VolumeGroupSnapshot.
+// If it is not set, gets it from default VolumeGroupSnapshotClass and sets it.
+// On error, it must return the original group snapshot, not nil, because the caller
+// syncGroupSnapshotByKey needs to check group snapshot's timestamp.
+func (ctrl *csiSnapshotCommonController) checkAndUpdateGroupSnapshotClass(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) (*crdv1alpha1.VolumeGroupSnapshot, error) {
+	className := groupSnapshot.Spec.VolumeGroupSnapshotClassName
+	var class *crdv1alpha1.VolumeGroupSnapshotClass
+	var err error
+	newGroupSnapshot := groupSnapshot
+	if className != nil {
+		klog.V(5).Infof("checkAndUpdateGroupSnapshotClass [%s]: VolumeGroupSnapshotClassName [%s]", groupSnapshot.Name, *className)
+		class, err = ctrl.getGroupSnapshotClass(*className)
+		if err != nil {
+			klog.Errorf("checkAndUpdateGroupSnapshotClass failed to getGroupSnapshotClass %v", err)
+			ctrl.updateGroupSnapshotErrorStatusWithEvent(groupSnapshot, false, v1.EventTypeWarning, "GetGroupSnapshotClassFailed", fmt.Sprintf("failed to get group snapshot class with error %v", err))
+			// we need to return the original group snapshot even if the class isn't found, as it may need to be deleted
+			return newGroupSnapshot, err
+		}
+	} else {
+		klog.V(5).Infof("checkAndUpdateGroupSnapshotClass [%s]: SetDefaultGroupSnapshotClass", groupSnapshot.Name)
+		class, newGroupSnapshot, err = ctrl.SetDefaultGroupSnapshotClass(groupSnapshot)
+		if err != nil {
+			klog.Errorf("checkAndUpdateGroupSnapshotClass failed to setDefaultClass %v", err)
+			ctrl.updateGroupSnapshotErrorStatusWithEvent(groupSnapshot, false, v1.EventTypeWarning, "SetDefaultGroupSnapshotClassFailed", fmt.Sprintf("Failed to set default group snapshot class with error %v", err))
+			return groupSnapshot, err
+		}
+	}
+
+	// For pre-provisioned group snapshots, we may not have group snapshot class
+	if class != nil {
+		klog.V(5).Infof("VolumeGroupSnapshotClass [%s] Driver [%s]", class.Name, class.Driver)
+	}
+	return newGroupSnapshot, nil
+}

pkg/utils/util.go

@@ -24,7 +24,6 @@ import (
 	"strings"
 	"time"
 
-	crdv1 "github.com/kubernetes-csi/external-snapshotter/client/v6/apis/volumesnapshot/v1"
 	v1 "k8s.io/api/core/v1"
 	"k8s.io/apimachinery/pkg/api/meta"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -33,6 +32,9 @@ import (
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/tools/cache"
 	klog "k8s.io/klog/v2"
+
+	crdv1alpha1 "github.com/kubernetes-csi/external-snapshotter/client/v6/apis/volumegroupsnapshot/v1alpha1"
+	crdv1 "github.com/kubernetes-csi/external-snapshotter/client/v6/apis/volumesnapshot/v1"
 )
 
 var keyFunc = cache.DeletionHandlingMetaNamespaceKeyFunc
@@ -475,3 +477,34 @@ func IsSnapshotReady(snapshot *crdv1.VolumeSnapshot) bool {
 func IsSnapshotCreated(snapshot *crdv1.VolumeSnapshot) bool {
 	return snapshot.Status != nil && snapshot.Status.CreationTime != nil
 }
+
+func GroupSnapshotKey(vgs *crdv1alpha1.VolumeGroupSnapshot) string {
+	return fmt.Sprintf("%s/%s", vgs.Namespace, vgs.Name)
+}
+
+func GroupSnapshotRefKey(vgsref *v1.ObjectReference) string {
+	return fmt.Sprintf("%s/%s", vgsref.Namespace, vgsref.Name)
+}
+
+func IsGroupSnapshotReady(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) bool {
+	if groupSnapshot.Status == nil || groupSnapshot.Status.ReadyToUse == nil || *groupSnapshot.Status.ReadyToUse == false {
+		return false
+	}
+	return true
+}
+
+func IsBoundVolumeGroupSnapshotContentNameSet(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot) bool {
+	if groupSnapshot.Status == nil || groupSnapshot.Status.BoundVolumeGroupSnapshotContentName == nil || *groupSnapshot.Status.BoundVolumeGroupSnapshotContentName == "" {
+		return false
+	}
+	return true
+}
+
+func IsVolumeGroupSnapshotRefSet(groupSnapshot *crdv1alpha1.VolumeGroupSnapshot, content *crdv1alpha1.VolumeGroupSnapshotContent) bool {
+	if content.Spec.VolumeGroupSnapshotRef.Name == groupSnapshot.Name &&
+		content.Spec.VolumeGroupSnapshotRef.Namespace == groupSnapshot.Namespace &&
+		content.Spec.VolumeGroupSnapshotRef.UID == groupSnapshot.UID {
+		return true
+	}
+	return false
+}