Recovering HDFS Service Data

Scenario

HDFS data needs to be recovered in the following scenarios: data is modified or deleted unexpectedly and needs to be restored. After a MRS cluster administrator performs critical data adjustment in HDFS, an exception occurs or the operation has not achieved the expected result. All modules are faulty and become unavailable. Data is migrated to a new cluster.

MRS cluster administrators can create a recovery task in FusionInsight Manager to recover HDFS data. Only manual recovery tasks are supported.

• Data recovery can be performed only when the system version is consistent with that of data backup.
• To recover data when the service is running properly, you are advised to manually back up the latest management data before recovering data. Otherwise, the HDFS data that is generated after the data backup and before the data recovery will be lost.
• The HDFS restoration operation cannot be performed for the directories used by running Yarn tasks, for example, /tmp/logs, /tmp/archived, and /tmp/hadoop-yarn/staging. Otherwise, data restoration using Distcp tasks fails due to file loss.
• If the backup data is EC data, data can only be restored to the directory configured with the same EC policy but not the directory not configured with the same EC policy or configured with the replica policy. Similarly, if the backup data is replica data, data can only be restored to the replica directory but not the EC directory.

Impact on the System

• During data recovery, user authentication stops and users cannot create new connections.
• After the data is recovered, the data produced between the backup time and restoration time is lost.
• After the data is recovered, the HDFS upper-layer applications need to be started.

Prerequisites

• To restore data from the remote HDFS, you need to prepare the standby cluster. If the active cluster is deployed in security mode and the active and standby clusters are not managed by the same FusionInsight Manager, configure system mutual trust. For details, see Configuring Cross-Manager Cluster Mutual Trust Relationships. If the active cluster is deployed in normal mode, do not configure mutual trust.

• Cross-cluster replication has been configured for the active and standby clusters. For details, see Enabling Cross-Cluster Replication.
• The time of the active cluster and standby cluster must be the same, and the NTP service in the active and standby clusters must use the same time source.
• The HDFS backup file save path is correct.
• Ensure that the backup data and target directory use the same EC policy or replica policy.
• The HDFS upper-layer applications are stopped.
• You have logged in to FusionInsight Manager. For details, see Logging In to FusionInsight Manager.

Procedure

1. On FusionInsight Manager, choose O&M > Backup and Restoration > Backup Management.
2. In the Operation column of a specified task in the task list, click More > View History to view historical backup task execution records.

   In the displayed window, locate a specified success record and click View in the Backup Path column to view the backup path information of the task and find the following information:

   • Backup Object specifies the data source of the backup data.
   • Backup Path specifies the full path where the backup files are saved.

     Select the correct item, and manually copy the full path of backup files in Backup Path.

3. On FusionInsight Manager, choose O&M > Backup and Restoration > Restoration Management.
4. Click Create.
5. Set Task Name to the name of the recovery task.
6. Select the cluster to be operated from Recovery Object.
7. Select HDFS.
8. Set Path Type of HDFS to a backup directory type.

   The following backup directory types are supported:

   • RemoteHDFS: indicates that the backup files are stored in the HDFS directory of the standby cluster.
     If you select this value, you need to set the following parameters:

     • Source NameService Name: indicates the NameService name of the backup data cluster. You can set it to the NameService name (haclusterX, haclusterX1, haclusterX2, haclusterX3, or haclusterX4) of the built-in remote cluster of the cluster, or the NameService name of a configured remote cluster.
     • IP Mode: mode of the target IP address.The system automatically selects the IP address mode based on the cluster network type, for example, IPv4 or IPv6.
     • Source NameNode IP Address: indicates the NameNode service plane IP address of the standby cluster, supporting the active node or standby node.

     • Source Path: indicates the full path of the HDFS directory for storing standby cluster backup data. For example, backup path/backup task name_data source_task creation time/.
     • Queue Name: indicates the name of the YARN queue used for backup task execution.
     • Recovery Point List: Click Refresh and select an HDFS directory that has been backed up in the standby cluster.
     • Target NameService Name: indicates the target NameService that corresponds to the selected backup directory. The default value of this parameter is hacluster.
     • Maximum Number of Maps: indicates the maximum number of maps in a MapReduce task. The default value of this parameter is 20.
     • Maximum Bandwidth of a Map (MB/s): indicates the maximum bandwidth of a map. The default value of this parameter is 100.
   • NFS: indicates that backup files are stored in the NAS over the NFS protocol.
     If you select NFS, set the following parameters:

     • IP Mode: mode of the target IP address.The system automatically selects the IP address mode based on the cluster network type, for example, IPv4 or IPv6.

     • Server IP address: indicates the NAS server IP address.
     • Source Path: indicates the complete path of the backup file on the NAS server. For example, backup path/backup task name_data source_task creation time/.
     • Queue Name: indicates the name of the YARN queue used for backup task execution.
     • Recovery Point List: Click Refresh and select an HDFS directory that has been backed up in the standby cluster.
     • Target NameService Name: indicates the target NameService that corresponds to the selected backup directory. The default value of this parameter is hacluster.
     • Maximum Number of Maps: indicates the maximum number of maps in a MapReduce task. The default value of this parameter is 20.
     • Maximum Bandwidth of a Map (MB/s): indicates the maximum bandwidth of a map. The default value of this parameter is 100.
   • CIFS: indicates that backup files are stored in the NAS over the CIFS protocol.
     If you select CIFS, set the following parameters:

     • IP Mode: mode of the target IP address.The system automatically selects the IP address mode based on the cluster network type, for example, IPv4 or IPv6.

     • Server IP address: indicates the NAS server IP address.
     • Port: indicates the port ID used by the CIFS protocol to connect to the NAS server. The default value is 445.
     • Username: Indicates the user name that is configured when setting the CIFS protocol.
     • Password: indicates the password that is configured when setting the CIFS protocol.
     • Source Path: indicates the complete path of the backup file on the NAS server. For example, backup path/backup task name_data source_task creation time/.
     • Queue Name: indicates the name of the YARN queue used for backup task execution.
     • Recovery Point List: Click Refresh and select an HDFS directory that has been backed up in the standby cluster.
     • Target NameService Name: indicates the target NameService that corresponds to the selected backup directory. The default value of this parameter is hacluster.
     • Maximum Number of Maps: indicates the maximum number of maps in a MapReduce task. The default value of this parameter is 20.
     • Maximum Bandwidth of a Map (MB/s): indicates the maximum bandwidth of a map. The default value of this parameter is 100.
   • SFTP: Indicates that backup files are stored in the server using SFTP.
     If you select SFTP, set the following parameters:

     • IP Mode: mode of the target IP address.The system automatically selects the IP address mode based on the cluster network type, for example, IPv4 or IPv6.

     • Server IP address: Enter the IP address of the server where the backup data is stored.
     • Port: Enter the port number used by the SFTP protocol to connect to the backup server. The default value is 22.
     • Username: Enter the username for connecting to the server using SFTP.
     • Password: Enter the password for connecting to the server using SFTP.
     • Source Path: Enter the full path of the backup file on the backup server. For example, backup path/backup task name_data source_task creation time/version_data source_task execution time.tar.gz.
     • Queue Name: indicates the name of the YARN queue used for backup task execution.
     • Recovery Point List: Click Refresh and select an HDFS directory that has been backed up in the standby cluster.
     • Target NameService Name: indicates the target NameService that corresponds to the selected backup directory. The default value of this parameter is hacluster.
     • Maximum Number of Maps: indicates the maximum number of maps in a MapReduce task. The default value of this parameter is 20.
     • Maximum Bandwidth of a Map (MB/s): indicates the maximum bandwidth of a map. The default value of this parameter is 100.

9. Set Backup Data column in Data Configuration to one or multiple backup data sources to be recovered based on service requirements. In the Target Path column, specify the target location after backup data recovery.

   You are advised to set Target Path to a new path that is difference from the backup path.

10. Click Verify to check whether the recovery task is configured correctly.
    • If the queue name is incorrect, the verification fails.
    • If the specified directory to be recovered does not exist, the verification fails.

11. Click OK to save the settings.
12. In the recovery task list, locate a created task and click Start in the Operation column to execute the recovery task.
    • After the recovery is successful, the progress bar is in green.
    • After the recovery is successful, the recovery task cannot be executed again.
    • If the recovery task fails during the first execution, rectify the fault and click Retry to execute the task again.