Restore from backup copy
A NetBackup restore from backup happens in parallel if there are multiple PVCs in the selected namespace. When you start a restore the job creates a parent-child hierarchy (if the namespace has at least one PVC to restore). The parent job acts as an orchestrator and monitors the status of child jobs. The first child job restores the metadata, after which PVCs are restored in parallel.
Note:
If metadata restore fails, no further jobs are submitted for restore operation. Once metadata is restored successfully, PVCs are restored parallel in batches.
You can follow the same procedure that is explained in restoring from snapshot, select the copy type as . You can also restore to alternate target cluster.
To restore from a backup copy
- On the left, click Workloads > Kubernetes.
- On the Namespace tab, click the namespace of the asset that you want to recover. Click the Recovery pointstab.
- The Recovery points tab shows you all the recovery points with the date, time, and copies of the backup. You can set filters to filter the displayed recovery points. Click the date in the Date column to view the details of the recovery point. The Recovery points details dialog shows the resources that were backed up, like
ConfigMaps, secrets, persistent volumes, pod, and so on. For details about these resources, see https://kubernetes.io/docs/reference. - Locate the recovery point that you want to restore.
- In the Copies column, click the # copies button. For example, if there are two copies, the button displays as 2 copies.
- In the list of copies, locate the Backup copy. Then click Actions > Restore.
Note:
Enable the Restore option for all infected copies by selecting the Allow the selection of recovery points that are malware-affected option.
- In the Recovery target page, to recover the asset to the same cluster source are auto populated. Click Next.
- Under Specify destination namespace, select from the options:
Select Use original namespace to use the original namespace.
Select Use alternate namespace and enter the alternate namespace.
Click Next.
- Under Select resource types to recover, select from the following resource types:
Select All resource types to recover all resource types.
Select Recover selected resource types to recover only the selected resource types.
- Under Select Persistent volume claims to recover, select from the following options:
Select All Persistent volume claims to recover all persistent volume claims.
Select Recover selected Persistent volume claims to recover selected persistent volume claims.
Note:
If you do not select any option in Recover selected resource types, then include empty persistent volume claims option is selected and no persistent volume claims are restored.
If you do not select any options in the Recover selected persistent volume claims, then in the Recovery options section it includes empty persistent volume claims and no persistent volume claims are restored.
Note:
Restore only persistent volume enables the toggle in the selected persistent volume claims to restore only the persistent volume. This setting does not create a corresponding persistent volume claim.
- Click on the Failure strategy section to view the failure strategy options to recover.
- Under Select failure strategy to recover, select any of the following failure strategies to recover:
Note:
On occurrence of any failure while restoring metadata or PVCs, a restore job runs as per the failure strategy selected.
Fail Fast to terminate the restore for any failure occurrence.
This restore failure strategy helps you to terminate the restore job when the first failure occurs.
All the remaining active restore jobs in the current batch are allowed to complete and no further batches are submitted for restore.
Proceed Ahead to continue restoring the next PVC. If the parent image (first image) restore fails, then the restore job terminates.
This strategy helps you to proceed ahead with the restore of remaining PVCs, if any of the PVC restores fail for the ongoing batch.
If metadata restore fails, the final job is marked as failed and no PVCs are submitted for restore.
In this case, the final job status which is marked as partial success and a list of PVCs with failed status appears in the Activity Monitor tab for the parent job.
Retry to specify a retry count for metadata or PVC restore. If the restore fails even after retries, then the restore job terminates.
This failure strategy helps you to retry the restore job of failed PVC/metadata which is configurable at the start of the restore job.
If the restore job fails despite the maximum number of retries, the job which is marked as failed and no further batches are submitted for restore.
Note:
The selected failure strategy is displayed in the Activity monitor.
Click Next.
- Click Start recovery to submit the recovery entry.
- In the Activity monitor, click the Job ID to view the restore job details.
- On the Job Details page, click the Details tab. The restore job sequence (pre-restore, data movement, and then the post-restore job) is displayed.
Note:
You can cancel the parent job to cancel the restore operation. The parent job terminates all the active child restore jobs.
The batch size for parallel PVC restore is configurable in bp.conf. User can add the key KUBERNETES_RESTORE_FROM_BACKUP_COPY_PARALLEL_RESTORE_BATCH_SIZE in bp.conf file to set the desired batch size. This key is optional and has the value of 5 if it is not defined.
The minimum value that can be assigned for batch size is 1, whereas the maximum value is 100.
You can use the bpsetconfig command on the NetBackup primary server to update the batch size.