Container Service for Kubernetes:Back up and restore cluster applications using kubectl

Scenarios

• You cannot use the backup center from the ACK console due to permission issues.

• You need to automate the disaster recovery and backup process.

Notes

• You cannot delete backup and restore jobs using the kubectl delete command. For more information, see Step 4: Delete resources related to the backup center.

• Monitor new releases of the migrate-controller component and update it to the latest version promptly. For more information, see Manage components.

• Do not delete the configuration parameters in the following examples. Deleting these parameters may cause the backup or restore to fail.

Step 1: Create a backup repository

1. In both the backup cluster and the restore cluster, create a file named backuplocation.yaml that contains the following content.

   apiVersion: csdr.alibabacloud.com/v1beta1
   kind: BackupLocation
   metadata:
     name: <yourBackuplocationName>
     namespace: csdr
   spec:
     backupSyncPeriod: 0s
     config:
       network: internal
       region: cn-beijing
     objectStorage:
       bucket: <cnfs-oss-yourBucketName>
       prefix: <subDir>
     provider: alibabacloud

   Retain the default configurations for the other parameters.

   Parameter	Required	Description
   name	Yes	The name of the backup repository. The name must follow Kubernetes naming conventions.
   network	Yes	The network access mode for the Object Storage Service (OSS) Bucket. Valid values: internal: The internal network access mode. All associated clusters and the OSS Bucket must be in the same region. public: The public network access mode. There are no region restrictions.
   region	Yes	The region where the OSS Bucket is located.
   bucket	Yes	The name of the OSS Bucket. You must create the bucket in advance. The name must start with cnfs-oss-****.
   prefix	No	The subdirectory in the OSS Bucket. If you set this parameter, the backup content is stored in this subdirectory.

2. Run the following command in both the backup cluster and the restore cluster to deploy the backuplocation object.

   kubectl apply -f backuplocation.yaml

3. Run the following command in both the backup cluster and the restore cluster to check the status of the backuplocation object.

   kubectl describe backuplocation <yourBackuplocationName> -n csdr

   Expected output:

   ...
   Status:
     Last Validation Time:  2022-12-08T04:00:22Z
     Message:               success by csdr-controller
     Phase:                 Available

   The expected output shows that the status of the backup repository is Available. This status indicates that the cluster has the permissions required to access OSS.

Step 2: Create a backup job

Step 3: Create a restore job

1. In the restore cluster, create a file named applicationrestore.yaml that contains the following content.

   apiVersion: csdr.alibabacloud.com/v1beta1
   kind: ApplicationRestore
   metadata:
     annotations:    
       csdr.alibabacloud.com/backuplocations: >-
         {"name":"<yourBackuplocationName>","region":"cn-beijing","bucket":"<cnfs-oss-yourBucketName>","prefix":"<subDir>","provider":"alibabacloud"}
     name: <yourApplicationRestoreName>
     namespace: csdr
   spec:
     appRestoreOnly: false
     preserveNodePorts: true
     includedNamespaces:
       - default
     includedResources:
       - statefulset
     excludedResources:
       - secret
     convertedarg:
     - convertToStorageClassType: alicloud-disk-topology-alltype
       convertToAccessModes:
       - ReadWriteOnce
       namespace: nas
       persistentVolumeClaim: pvc-nas
     - convertToStorageClassType: alicloud-disk-topology-alltype
       namespace: oss
       persistentVolumeClaim: pvc-oss
     backupName: <yourApplicationBackupName>
     namespaceMapping:
       <backupNamespace>: <restoreNamespace>
     imageRegistryMapping:
       <oldImageRegistry>: <newImageRegistry>

   Retain the default configurations for the other parameters.

   Note

   The appRestoreOnly, preserveNodePorts, includedResources, and excludedResources parameters apply only to application backups. These parameters are ignored in data protection scenarios.

   Parameter	Required	Description
   csdr.alibabacloud.com/backuplocations	Yes	Information about the backup repository where the backup is stored. This must be consistent with the backup repository configuration.
   name	Yes	The name of the restore job.
   appRestoreOnly	No	This parameter is valid only for application backup jobs. It does not take effect for data protection jobs. For backups that include PVs, specifies whether to restore only the application and not the PVCs, PVs, and their corresponding data. Valid values: true: Restores only the application. To change the data source of the backed-up application, you can manually create the PVCs and PVs, then create a restore job and set this parameter to true. false: Restores the application and the related storage data. This is the default value.
   preserveNodePorts	No	This parameter is valid only for application backup jobs. It does not take effect for data protection jobs. Specifies whether to keep the NodePort of the application. When the backup cluster and the restore cluster are the same, the default value is false. The port value is randomly changed to avoid conflicts with existing applications of the same type. Otherwise, set this to true.
   includedNamespaces	Yes	Specifies the name of the namespace to restore. If you leave this empty, all backed-up namespaces are restored.
   includedResources	No	This parameter is valid only for application backup jobs. It does not take effect for data protection jobs. Specifies the cluster resource types to restore. Important To avoid unexpected results, configure only one of includedResources or excludedResources. If both are empty, all backed-up resource types are restored.
   excludedResources	No	This parameter is valid only for application backup jobs. It does not take effect for data protection jobs. Excludes the cluster resource types to restore. Important To avoid unexpected results, configure only one of includedResources or excludedResources. If both are empty, all backed-up resource types are restored.
   backupName	Yes	The name of the backup to restore. If you use the scheduled backup feature, you must specify the name of a backup created at a specific point in time, such as <yourBackupScheduleName>-20221205225845.
   namespaceMapping	No	Remaps the namespace. Valid values: <backupNamespace>: The namespace in the backup cluster that was specified when the backup job was created. <restoreNamespace>: The namespace in the restore cluster where the resources from the backup are created. If you do not set this, the resources are restored to a namespace with the same name by default. Note If <restoreNamespace> does not exist, a new namespace is created.
   imageRegistryMapping	No	Remaps the image repository address. This applies to the image configurations of all applications that match the rule. Valid values: <oldImageRegistry>: The original image repository address of the application in the backup. <newImageRegistry>: The target image repository address for the restore. If you do not set this, the image configuration in the application is not changed. Example: The original application uses the image registry.cn-beijing.aliyuncs.com/my-registry/nginx:v1. You need to change it to registry.cn-hangzhou.aliyuncs.com/my-registry/nginx:v1. Configure the field as follows: imageRegistryMapping: "registry.cn-beijing.aliyuncs.com/my-registry/": "registry.cn-hangzhou.aliyuncs.com/my-registry/"
   convertedarg	No	A list of StorageClass transformations. The PVCs of FileSystem data type volumes (such as OSS, NAS, CPFS, and local storage) in the backup are restored to the target StorageClass in the current cluster. This allows transformations, for example, from NAS storage to disk storage. convertToStorageClassType: The target StorageClass. It must exist in the current cluster. The current version only supports selecting a disk or NAS StorageClass. namespace: The namespace where the storage claim is located. persistentVolumeClaim: The name of the PVC. The preceding parameters are required for the StorageClass transformation feature. When transforming a StorageClass, you can also modify the AccessModes configuration of the PV using the following optional parameter. convertToAccessModes: A list of target AccessModes. Note When a PV with an AccessModes of ReadWriteMany or ReadOnlyMany needs to be restored to a disk StorageClass, you must also transform the AccessModes to ReadWriteOnce to avoid the risk of forced disk detachment. You can run kubectl -ncsdr describe <backup-name> to query the PVC information of the backup. In the status.resourceList.dataResource.pvcBackupInfo list of the output, the dataType field indicates the data type, which can be FileSystem or Snapshot. The nameSpace and pvcName fields are the namespace and name of the PVC, respectively.

2. Run the following command in the restore cluster to deploy the applicationrestore object.

   kubectl apply -f applicationrestore.yaml

3. Run the following command in the restore cluster to check the status of the restore job.

   kubectl describe applicationrestore <yourApplicationRestoreName> -n csdr

   Expected output:

   ...
   Status:
     Completion Timestamp:  2022-12-05T15:52:19Z
     Phase:                 Completed
     Start Timestamp:       2022-12-05T15:52:09Z

   The restore job is complete when its status changes from Inprogress to Completed.

Step 4: Delete resources related to the backup center

Delete a backup schedule

Run the following command to delete the BackupSchedule resource and stop scheduled backups.

kubectl delete backupschedule <yourBackupSchedule> -n csdr

Delete a backup or restore job

1. In the cluster that contains the job you want to delete, create a file named deleterequest.yaml that contains the following content.

   apiVersion: csdr.alibabacloud.com/v1beta1
   kind: DeleteRequest
   metadata:
     name: <objectName-dbr>
     namespace: csdr
   spec:
     deleteObjectName: <objectName>
     deleteObjectType: "Backup"

   Retain the default configurations for the other parameters.

   Parameter	Required	Description
   name	Yes	The name of the delete request. For a backup job: The name of the ApplicationBackup to be deleted, plus "-dbr". For a restore job: The name of the ApplicationRestore to be deleted, plus "-dbr".
   deleteObjectName	Yes	The name of the resource to delete.
   deleteObjectType	Yes	The type of resource to delete. Valid values: "Backup": Deletes the ApplicationBackup job and its associated resources. "Restore": Deletes the ApplicationRestore job and its associated resources. Note Deleting a backup job does not affect replicas that have been synchronized to the cluster. Deleting a restore job does not affect restored content.

2. Run the following command in the cluster that contains the job you want to delete to deploy the deleterequest object.

   kubectl apply -f deleterequest.yaml

3. Use the following commands to check the resource status.

   • Check the deleted backup job

     • Run the following command to check the status of the backup job deletion.

       kubectl get applicationbackup <yourApplicationBackupName> -n csdr

       Expected output:

       Error from server (NotFound): applicationbackups.csdr.alibabacloud.com "yourApplicationBackupName" not found

       An empty output indicates that the backup job resource has been deleted.

     • Run the following command to check the status of the delete request resource.

       kubectl get deleterequest <yourApplicationBackupName-dbr> -n csdr

       Expected output:

       Error from server (NotFound): deleterequests.csdr.alibabacloud.com "yourApplicationBackupName-dbr" not found

       The output shows that the backup job resource and its related delete request resources are automatically deleted.

   • Check the deleted restore job

     • Run the following command to check the status of the deleted restore job.

       kubectl get applicationrestore <yourApplicationRestoreName> -n csdr

       Expected output:

       Error from server (NotFound): applicationrestores.csdr.alibabacloud.com "yourApplicationRestoreName" not found

       An empty output indicates that the restore job resource has been deleted.

     • Run the following command to check the status of the delete request resource.

       kubectl get deleterequest <yourApplicationRestoreName-dbr> -n csdr

       Expected output:

       Error from server (NotFound): deleterequests.csdr.alibabacloud.com "yourApplicationRestoreName-dbr" not found

       An empty output indicates that the related delete request resource is also automatically deleted.

References

• To back up and migrate cluster applications from the console, see the following documents:

  • Back up and restore applications within a cluster

  • Migrate applications across clusters in the same region

  • Migrate applications across clusters in different regions

• To migrate applications between clusters with different storage plug-ins or Kubernetes versions, see Migrate applications from a low-version Kubernetes cluster using the backup center.

• Back up and restore applications using Terraform