fix backup stage order

cmd/backup-manager/app/backup/manager.go

@@ -142,13 +142,6 @@ func (bm *Manager) performBackup(ctx context.Context, backup *v1alpha1.Backup, d
 	started := time.Now()
 
 	var errs []error
-	err := bm.StatusUpdater.Update(backup, &v1alpha1.BackupCondition{
-		Type:   v1alpha1.BackupRunning,
-		Status: corev1.ConditionTrue,
-	}, nil)
-	if err != nil {
-		return err
-	}
 
 	backupFullPath, err := util.GetStoragePath(backup)
 	if err != nil {
@@ -166,11 +159,10 @@ func (bm *Manager) performBackup(ctx context.Context, backup *v1alpha1.Backup, d
 	updatePathStatus := &controller.BackupUpdateStatus{
 		BackupPath: &backupFullPath,
 	}
-	err = bm.StatusUpdater.Update(backup, &v1alpha1.BackupCondition{
+	if err := bm.StatusUpdater.Update(backup, &v1alpha1.BackupCondition{
 		Type:   v1alpha1.BackupPrepare,
 		Status: corev1.ConditionTrue,
-	}, updatePathStatus)
-	if err != nil {
+	}, updatePathStatus); err != nil {
 		return err
 	}
 
@@ -260,6 +252,14 @@ func (bm *Manager) performBackup(ctx context.Context, backup *v1alpha1.Backup, d
 		}
 	}
 
+	// change Prepare to Running before real backup process start
+	if err := bm.StatusUpdater.Update(backup, &v1alpha1.BackupCondition{
+		Type:   v1alpha1.BackupRunning,
+		Status: corev1.ConditionTrue,
+	}, nil); err != nil {
+		return err
+	}
+
 	// run br binary to do the real job
 	backupErr := bm.backupData(ctx, backup)
 

cmd/backup-manager/app/export/export.go

@@ -59,11 +59,11 @@ func (bo *Options) getDestBucketURI(remotePath string) string {
 	return fmt.Sprintf("%s://%s", bo.StorageType, remotePath)
 }
 
-func (bo *Options) dumpTidbClusterData(ctx context.Context, backup *v1alpha1.Backup) (string, error) {
+func (bo *Options) dumpTidbClusterData(ctx context.Context, backup *v1alpha1.Backup) error {
 	bfPath := bo.getBackupFullPath()
 	err := backupUtil.EnsureDirectoryExist(bfPath)
 	if err != nil {
-		return "", err
+		return err
 	}
 	args := []string{
 		fmt.Sprintf("--output=%s", bfPath),
@@ -88,9 +88,9 @@ func (bo *Options) dumpTidbClusterData(ctx context.Context, backup *v1alpha1.Bac
 
 	output, err := exec.CommandContext(ctx, binPath, args...).CombinedOutput()
 	if err != nil {
-		return bfPath, fmt.Errorf("cluster %s, execute dumpling command %v failed, output: %s, err: %v", bo, args, string(output), err)
+		return fmt.Errorf("cluster %s, execute dumpling command %v failed, output: %s, err: %v", bo, args, string(output), err)
 	}
-	return bfPath, nil
+	return nil
 }
 
 func (bo *Options) backupDataToRemote(ctx context.Context, source, bucketURI string, opts []string) error {

cmd/backup-manager/app/export/manager.go

@@ -154,7 +154,7 @@ func (bm *BackupManager) performBackup(ctx context.Context, backup *v1alpha1.Bac
 	started := time.Now()
 
 	err := bm.StatusUpdater.Update(backup, &v1alpha1.BackupCondition{
-		Type:   v1alpha1.BackupRunning,
+		Type:   v1alpha1.BackupPrepare,
 		Status: corev1.ConditionTrue,
 	}, nil)
 	if err != nil {
@@ -242,7 +242,23 @@ func (bm *BackupManager) performBackup(ctx context.Context, backup *v1alpha1.Bac
 		klog.Infof("set cluster %s %s to %s success", bm, constants.TikvGCVariable, constants.TikvGCLifeTime)
 	}
 
-	backupFullPath, backupErr := bm.dumpTidbClusterData(ctx, backup)
+	backupFullPath := bm.getBackupFullPath()
+	// TODO: Concurrent get file size and upload backup data to speed up processing time
+	archiveBackupPath := backupFullPath + constants.DefaultArchiveExtention
+	remotePath := strings.TrimPrefix(archiveBackupPath, constants.BackupRootPath+"/")
+	bucketURI := bm.getDestBucketURI(remotePath)
+	updatePathStatus := &controller.BackupUpdateStatus{
+		BackupPath: &bucketURI,
+	}
+	err = bm.StatusUpdater.Update(backup, &v1alpha1.BackupCondition{
+		Type:   v1alpha1.BackupRunning,
+		Status: corev1.ConditionTrue,
+	}, updatePathStatus)
+	if err != nil {
+		return err
+	}
+
+	backupErr := bm.dumpTidbClusterData(ctx, backup)
 	if oldTikvGCTimeDuration < tikvGCTimeDuration {
 		// use another context to revert `tikv_gc_life_time` back.
 		// `DefaultTerminationGracePeriodSeconds` for a pod is 30, so we use a smaller timeout value here.
@@ -296,8 +312,6 @@ func (bm *BackupManager) performBackup(ctx context.Context, backup *v1alpha1.Bac
 	}
 	klog.Infof("get cluster %s commitTs %s success", bm, commitTs)
 
-	// TODO: Concurrent get file size and upload backup data to speed up processing time
-	archiveBackupPath := backupFullPath + constants.DefaultArchiveExtention
 	err = archiveBackupData(backupFullPath, archiveBackupPath)
 	if err != nil {
 		errs = append(errs, err)
@@ -332,19 +346,6 @@ func (bm *BackupManager) performBackup(ctx context.Context, backup *v1alpha1.Bac
 	// archive backup data successfully, origin dir can be deleted safely
 	os.RemoveAll(backupFullPath)
 
-	remotePath := strings.TrimPrefix(archiveBackupPath, constants.BackupRootPath+"/")
-	bucketURI := bm.getDestBucketURI(remotePath)
-	updatePathStatus := &controller.BackupUpdateStatus{
-		BackupPath: &bucketURI,
-	}
-	err = bm.StatusUpdater.Update(backup, &v1alpha1.BackupCondition{
-		Type:   v1alpha1.BackupPrepare,
-		Status: corev1.ConditionTrue,
-	}, updatePathStatus)
-	if err != nil {
-		return err
-	}
-
 	err = bm.backupDataToRemote(ctx, archiveBackupPath, bucketURI, opts)
 	if err != nil {
 		errs = append(errs, err)