Initiate an incremental data backup job

An incremental data backup job backs up all macroblocks modified since the last full backup or incremental data backup. For example, when the system initiates an incremental data backup job for a tenant, if a full backup was performed in the last data backup job and a full backup file named full_backup_set was generated, the system will back up all macroblocks modified since the generation of full_backup_set, and if an incremental backup was performed in the last data backup job and an incremental backup file named inc_backup_set was generated, the system will back up all macroblocks modified since the generation of inc_backup_set.

Limitations and considerations

• To perform an incremental data backup, make sure that a full data backup exists. If an incremental backup is initiated without a full data backup, the system will automatically convert the incremental backup to a full backup.

• For scenarios where a cluster of a low version is upgraded to a high version, even if the current tenant has performed a full backup in the low version, when upgrading to the high version, if an incremental backup is directly initiated, the system will automatically convert the incremental backup to a full backup.

Procedure

After you complete the preparations, you can perform the following steps to initiate an incremental data backup job.

Assume that the current cluster contains three tenants: sys, mysql_tenant, and oracle_tenant.

Initiate an incremental data backup job from the sys tenant

You can initiate an incremental data backup job for all tenants or a specified tenant in the cluster from the sys tenant.

1. Log in to the sys tenant of the cluster as the root user.

2. (Optional) Set the backup concurrency for the tenant.

   You can configure the backup concurrency by setting the tenant-level parameter ha_low_thread_score, which specifies the number of worker threads for job queues with a low or medium priority, such as backup job queues and backup cleanup job queues. The value range is [0, 100]. The default value is 0. Before you start the backup job, you can increase the value of ha_low_thread_score. We recommend that you double the value each time. For more information about the operations related to the ha_low_thread_score parameter, see View data backup parameter settings.

   For more information about the ha_low_thread_score parameter, see ha_low_thread_score.

   • Execute the following statement to set the backup concurrency for all tenants in the cluster:

     obclient [(none)]>  ALTER SYSTEM SET ha_low_thread_score = 10 TENANT = all_user;
     

     Alternatively, execute the following statement:

     obclient [(none)]>  ALTER SYSTEM SET ha_low_thread_score = 10 TENANT = all;
     

     Note

     In OceanBase Database V4.2.1 and later, TENANT = all_user and TENANT = all express the same semantics. If you want an operation to take effect on all user tenants, we recommend that you use TENANT = all_user. TENANT = all will be deprecated.

   • Execute the following statement to set the backup concurrency for a specified tenant in the cluster:

     obclient [(none)]>  ALTER SYSTEM SET ha_low_thread_score = 10 TENANT = mysql_tenant;
     

3. Initiate an incremental data backup job.

   • Initiate an incremental data backup job for all tenants in the cluster

     Execute the following statement to initiate an incremental data backup job for all tenants in the cluster.

     obclient [(none)]> ALTER SYSTEM BACKUP INCREMENTAL DATABASE;
     

     After the statement is executed, the system initiates an incremental data backup job for the mysql_tenant and oracle_tenant tenants in the cluster.

   • Initiate an incremental data backup job for a specified tenant in the cluster

     You can initiate an incremental data backup job for a specified tenant without affecting other tenants in the cluster.

     To initiate an incremental data backup job for the mysql_tenant tenant, execute the following statement:

     obclient [(none)]> ALTER SYSTEM BACKUP INCREMENTAL TENANT = mysql_tenant;
     

     Note

     If you specify multiple tenants, separate the tenant names with commas (,).

     After the statement is executed, the system initiates an incremental data backup job for the mysql_tenant tenant.

   You can view the backup progress in real time. For more information, see View the data backup progress.

Initiate an incremental data backup job for a user tenant

You can initiate an incremental data backup job for a user tenant without affecting other tenants.

1. Log in to the database as a tenant administrator.

   In this example, you can log in to the mysql_tenant tenant as the root user, or to the oracle_tenant tenant as the SYS user.

2. (Optional) Set the backup concurrency.

   You can configure the backup concurrency by setting the tenant-level parameter ha_low_thread_score, which specifies the number of worker threads for job queues with a low or medium priority, such as backup job queues and backup cleanup job queues. The value range is [0, 100]. The default value is 0. Before you start the backup job, you can increase the value of ha_low_thread_score. We recommend that you double the value each time. For more information about the operations related to the ha_low_thread_score parameter, see View data backup parameter settings.

   For more information about the ha_low_thread_score parameter, see ha_low_thread_score.

   obclient [(none)]>  ALTER SYSTEM SET ha_low_thread_score = 10;
   

3. Execute the following statement to initiate an incremental data backup job:

   obclient [(none)]> ALTER SYSTEM BACKUP INCREMENTAL DATABASE;
   

   After the statement is executed, the system initiates an incremental data backup job for the current tenant.

   You can view the backup progress in real time. For more information, see View the data backup progress.

References

• View the data backup progress

• View the data backup history