Added a CLI command to take a single DB backup

Signed-off-by: Danny Zaken <dannyzaken@gmail.com>

Author: dannyzaken

pkg/cnpg/cnpg.go

@@ -620,6 +620,7 @@ func GetCnpgBackupObj(namespace string, name string) *cnpgv1.Backup {
 			Name:      name,
 			Namespace: namespace,
 		},
+		Spec: cnpgv1.BackupSpec{},
 	}
 	return cnpgBackup
 }

pkg/system/db_reconciler.go

@@ -32,16 +32,14 @@ const (
 
 // ReconcileCNPGCluster reconciles the CNPG cluster
 // There are several cases to handle:
-// 1. Reconciling a fresh install - No CNPG cluster and no previous DB to import from
+// 1. Reconciling a fresh install - No CNPG cluster and no DBRecovery configuration
 //   - In this case we need to create a new empty CNPG cluster
 //   - Create a new CNPG image catalog
 //   - Create a new CNPG cluster
 //
-// 2. Reconciling an upgrade from a version with a standalone DB - No CNPG cluster and DB statefulset exists
-//   - In this case we need to create a new CNPG cluster and import the DB from the previous statefulset
-//   - Import is done by providing externalCluster details in the CNPG cluster spec (https://cloudnative-pg.io/documentation/1.25/database_import/#the-microservice-type)
-//   - After Import is completed, cleanup the old DB resources. For now we only scale down the standalone DB pod to 0 replicas
-//   - All other pods (core, endpoints) should be stopped before starting the import
+// 2. Reconciling recovery from a snapshot - CNPG cluster was deleted by the user to initiate a recovery from a snapshot
+//   - In this case we need to create a new CNPG cluster and set the bootstrap configuration to recover from the snapshot
+//   - All other pods (core, endpoints) should be stopped before starting the recovery
 //
 // 3. Reconciling an existing CNPG cluster with no standalone DB - CNPG cluster exists and DB statefulset does not exist
 //   - If the major version is the same, check if the DB image is changed and update the ImageCatalog
@@ -149,8 +147,11 @@ func (r *Reconciler) reconcileDBCluster() error {
 
 	// Apply changes to the cluster resources. create or update the modified cluster
 	if r.CNPGCluster.UID == "" {
+		// If the cluster CR has not been created yet (UID = ""), it means one of two options.:
+		// 1. The cluster is being created for the first time
+		// 2. The cluster is being recovered from a snapshot
 
-		// Cluster resource is missing. Check if noobaa CR has a recovery configuration.
+		// Check if noobaa CR has a recovery configuration.
 		if r.NooBaa.Spec.DBSpec.DBRecovery == nil {
 			// No recovery configuration found, set bootstrap configuration to init a new DB
 			if r.CNPGCluster.Spec.Bootstrap == nil {
@@ -205,11 +206,11 @@ func (r *Reconciler) reconcileDBCluster() error {
 			r.cnpgLog("cluster spec is changed, updating cluster. diff: %v", diff)
 
 			currentDBClusterStatus := r.NooBaa.Status.DBStatus.DBClusterStatus
-			// avoid updating a cluster that is being created or imported.
+			// avoid updating a cluster that is being created.
 			// We might want to consider allowing this somehow for supportability (through annotation or something)
-			if currentDBClusterStatus == nbv1.DBClusterStatusCreating || currentDBClusterStatus == nbv1.DBClusterStatusImporting {
-				r.cnpgLog("the cluster spec was changed but the cluster creation or import is still in progress, skipping update")
-				return fmt.Errorf("cluster creation or import is still in progress, skipping update")
+			if currentDBClusterStatus == nbv1.DBClusterStatusCreating || currentDBClusterStatus == nbv1.DBClusterStatusRecovering {
+				r.cnpgLog("the cluster spec was changed but the cluster creation is still in progress, skipping update")
+				return fmt.Errorf("cluster creation is still in progress, skipping update")
 			}
 
 			r.cnpgLog("cluster spec is changed, updating cluster")
@@ -444,6 +445,7 @@ func (r *Reconciler) reconcileScheduledBackup() error {
 		scheduledBackup.Spec.Cluster.Name = r.CNPGCluster.Name
 		scheduledBackup.Spec.Method = cnpgv1.BackupMethodVolumeSnapshot
 		scheduledBackup.Spec.Online = &offlineBackup
+		scheduledBackup.Spec.Target = cnpgv1.BackupTargetStandby
 		if scheduledBackup.Status.LastScheduleTime != nil {
 			r.NooBaa.Status.DBStatus.BackupStatus.LastBackupTime = scheduledBackup.Status.LastScheduleTime
 		}

pkg/system/system.go

@@ -12,6 +12,8 @@ import (
 	"strings"
 	"time"
 
+	cnpgv1 "github.com/cloudnative-pg/cloudnative-pg/api/v1"
+	storagesnapshotv1 "github.com/kubernetes-csi/external-snapshotter/client/v8/apis/volumesnapshot/v1"
 	nbv1 "github.com/noobaa/noobaa-operator/v5/pkg/apis/noobaa/v1alpha1"
 	"github.com/noobaa/noobaa-operator/v5/pkg/bundle"
 	"github.com/noobaa/noobaa-operator/v5/pkg/cnpg"
@@ -47,6 +49,7 @@ func Cmd() *cobra.Command {
 		CmdCreate(),
 		CmdDelete(),
 		CmdStatus(),
+		CmdDBBackup(),
 		CmdSetDebugLevel(),
 		CmdList(),
 		CmdReconcile(),
@@ -118,6 +121,18 @@ func CmdSetDebugLevel() *cobra.Command {
 	return cmd
 }
 
+// CmdDBBackup returns a CLI command
+func CmdDBBackup() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "db-backup",
+		Short: "Create a volume snapshot backup of the database",
+		Run:   RunDBBackup,
+		Args:  cobra.NoArgs,
+	}
+	cmd.Flags().String("name", "", "A custom name for the backup")
+	return cmd
+}
+
 // CmdReconcile returns a CLI command
 func CmdReconcile() *cobra.Command {
 	cmd := &cobra.Command{
@@ -891,6 +906,85 @@ func RunSetDebugLevel(cmd *cobra.Command, args []string) {
 	fmt.Println("Debug level is not persistent and is only effective for the currently running core and endpoints pods")
 }
 
+// RunDBBackup runs a CLI command
+func RunDBBackup(cmd *cobra.Command, args []string) {
+	log := util.Logger()
+
+	sys := &nbv1.NooBaa{
+		TypeMeta: metav1.TypeMeta{Kind: "NooBaa"},
+		ObjectMeta: metav1.ObjectMeta{
+			Name:      options.SystemName,
+			Namespace: options.Namespace,
+		},
+	}
+
+	if !util.KubeCheck(sys) {
+		log.Fatalf("❌ System %q not found", options.SystemName)
+	}
+
+	if sys.Spec.DBSpec == nil {
+		log.Fatalf("❌ The system is not configured with a CNPG cluster")
+	}
+
+	if sys.Spec.DBSpec.DBBackup == nil || sys.Spec.DBSpec.DBBackup.VolumeSnapshot == nil {
+		log.Fatalf("❌ The system is not configured with a volume snapshot backup")
+	}
+
+	volumeSnapshotClass := sys.Spec.DBSpec.DBBackup.VolumeSnapshot.VolumeSnapshotClass
+	if volumeSnapshotClass == "" {
+		log.Fatalf("❌ The system is not configured with a volume snapshot class")
+	}
+
+	backupName, _ := cmd.Flags().GetString("name")
+	if backupName == "" {
+		backupName = sys.Name + pgClusterSuffix + "-backup-" + time.Now().Format("20060102150405")
+	}
+
+	offlineBackup := false
+	backup := cnpg.GetCnpgBackupObj(sys.Namespace, backupName)
+	backup.Spec.Cluster.Name = sys.Name + pgClusterSuffix
+	backup.Spec.Method = cnpgv1.BackupMethodVolumeSnapshot
+	backup.Spec.Online = &offlineBackup
+	backup.Spec.Target = cnpgv1.BackupTargetStandby
+
+	if !util.KubeCreateFailExisting(backup) {
+		log.Fatalf("❌ Backup %s failed to create", backupName)
+	}
+
+	log.Printf("✅ Backup object %s created successfully. Waiting for the volume snapshot to be created...\n", backupName)
+	log.Printf("You can monitor the backup status with the following command:\n")
+	log.Printf("kubectl -n %s get backups.postgresql.cnpg.noobaa.io %s", sys.Namespace, backupName)
+
+	// Create a context with timeout for polling
+	pollTimeout := 5 * time.Minute
+	pollCtx, cancel := context.WithTimeout(context.Background(), pollTimeout)
+	defer cancel()
+
+	// wait for the volume snapshot to be created
+	interval := time.Duration(3)
+	err := wait.PollUntilContextCancel(pollCtx, interval*time.Second, true, func(ctx context.Context) (bool, error) {
+		volumeSnapshot := &storagesnapshotv1.VolumeSnapshot{}
+		volumeSnapshotErr := util.KubeClient().Get(util.Context(),
+			client.ObjectKey{Namespace: sys.Namespace, Name: backupName},
+			volumeSnapshot)
+		if errors.IsNotFound(volumeSnapshotErr) {
+			log.Printf("⏳ Volume snapshot %s not found yet", backupName)
+			return false, nil
+		}
+		if volumeSnapshotErr != nil {
+			return false, volumeSnapshotErr
+		}
+		return true, nil
+	})
+	if err != nil {
+		log.Fatalf("❌ Failed to wait for volume snapshot %s: %s", backupName, err)
+	}
+
+	log.Printf("✅ Volume snapshot %s created successfully\n", backupName)
+	log.Printf("You can view the volume snapshot with the following command:\n")
+	log.Printf("kubectl -n %s get volumesnapshots.snapshot.storage.k8s.io %s", sys.Namespace, backupName)
+}
+
 // RunReconcile runs a CLI command
 func RunReconcile(cmd *cobra.Command, args []string) {
 	log := util.Logger()