After the multiversion function is enabled, expired backup data will be directly deleted by the system when the insert overwrite or truncate statement is executed. You can configure the trash bin of the OBS parallel file system to accelerate the deletion of expired backup data. To enable the trash bin, add dli.multi.version.trash.dir to the table properties. For details about the syntax for enabling or disabling the multiversion function, see Enabling or Disabling Multiversion Backup.
Currently, the multiversion function supports only OBS tables created using the Hive syntax. For details about the syntax for creating a table, see Creating an OBS Table Using the Hive Syntax.
1 2 | ALTER TABLE [db_name.]table_name SET TBLPROPERTIES ("dli.multi.version.trash.dir"="OBS bucket for expired multiversion backup data"); |
Parameter |
Description |
---|---|
db_name |
Database name, which consists of letters, digits, and underscores (_). The value cannot contain only digits or start with a digit or underscore (_). |
table_name |
Table name |
OBS bucket for expired multiversion backup data |
A directory in the bucket where the current OBS table locates. You can change the directory path as needed. For example, if the current OBS table directory is obs://bucketName/filePath and a Trash directory has been created in the OBS table directory, you can set the trash bin directory to obs://bucketName/filePath/Trash. |
1 2 | ALTER TABLE test_table SET TBLPROPERTIES ("dli.multi.version.trash.dir"="/.Trash"); |