Configure dynamic expansion of disk data files

OceanBase Database supports automatic expansion of disk data files based on their actual usage.

Background information

In OceanBase Database, the file that stores data is named block_file. The file is created after each OBServer node starts up and is stored in the /OBServer installation directory/store/sstable directory. The size of the block_file file is controlled by the datafile_size or datafile_disk_percentage configuration parameter. For more information about the data files of OceanBase Database, see Structure of the OBServer node installation directory.

For OceanBase Database V4.2.0 and earlier, the system mainly pre-allocates disk space for data files. This ensures data files occupy a relatively contiguous segment of disk space, avoiding resource contention from other applications. However, this approach reserves significant disk space, which cannot be reclaimed even if not used.

To resolve the issue of pre-allocating large disk spaces that cannot be released, starting from OceanBase Database V4.2.0, a user-configurable option for data files to incrementally use disk space has been introduced. This means the system will first allocate a reasonable disk size to data files, and then automatically expand as needed based on actual disk usage and user configuration.

Related Parameters

OceanBase Database mainly uses the following configuration items to automatically expand disk data files:

• datafile_size : the maximum available disk space that can be occupied by the data files. The default value is 0M. For more information about the cluster-level parameter datafile_size, see datafile_size.

• datafile_disk_percentage: the percentage of disk space occupied by the data file of a node. The default value is 0. For more information about the cluster-level parameter datafile_disk_percentage, see datafile_disk_percentage.

• datafile_next: specifies the incremental step for autoextending disk data files. The default value is 0, indicating that autoextension is disabled. To enable autoextension, set this to a non-zero value. For more information about the cluster-level configuration parameter datafile_next, see datafile_next.

• datafile_maxsize: specifies the maximum storage space allowed for automatic expansion of data files. The default value is 0, which disables automatic expansion. For more information about the cluster-level configuration parameter datafile_maxsize, see datafile_maxsize.

When you configure the preceding parameters, note the following:

• You can use either datafile_size or datafile_disk_percentage to set the data file disk space. In other cases:

  • If both datafile_size and datafile_disk_percentage are specified as non-zero values, the value of datafile_size takes precedence.

  • If neither datafile_size nor datafile_disk_percentage is configured (i.e., datafile_size is set to OM and datafile_disk_percentage is set to 0), the system will automatically calculate the percentage of disk space occupied by data files, based on whether logs and data share the same disk:

    • If the disk used by logs and data is shared, data files occupy 60% of the total space available on the disk.

    • If the logs and data are stored on separate disks, then the data files use 90% of the disk space on which they reside.

• When configuring datafile_maxsize, the value must be greater than the current disk space occupied by the data file datafile_size (or datafile_disk_percentage). If the set value is less than the current disk space occupied by the data file, automatic expansion will not be triggered.

• The maximum size of a data file is the smaller value of the datafile_maxsize and the maximum available space on the disk.

• For datafile_next:

  • If datafile_next is set to a value less than or equal to 1G, then datafile_next will be set to min (1G, datafile_maxsize * 10%).

  • When datafile_next is set to a value greater than 1G, the value of datafile_next is min ( datafile_next, disk space available)

Limitations and considerations

• Configure the initial value of the step size datafile_next as approximately 20% of datafile_maxsize to reduce frequent expansions when configuring automatic disk data file expansion.

• After data file auto-expansion is enabled, ensure that the maximum size to which a data file can expand is not less than the value specified for datafile_maxsize for the applications deployed on the same host.

• Dynamic disk data file shrinking is not supported.

• If you do not want to enable automatic expansion, you can directly set the value of datafile_next or datafile_maxsize to 0M or leave it at its default value of 0. The system will then allocate a portion of disk space to data files using the original pre-allocation method.

Enable automatic expansion of data files

Step 1: View the values of the current configuration items

Before configuring dynamic expansion of disk data files, you need to view the values of the configuration items to understand their current settings.

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

2. Execute the following command to view the values of the configuration items.

   obclient> SHOW PARAMETERS LIKE '%datafile_%';
   

Step 2: Configure dynamic expansion of disk data files

After obtaining the values of the configuration items, you can modify the values of the configuration items based on your actual usage scenarios to enable automatic expansion of disk data files.

If datafile_size (or datafile_disk_percentage), datafile_next, and datafile_maxsize are configured to non-zero values when the cluster is started

If you have configured datafile_size (or datafile_disk_percentage), datafile_next, and datafile_maxsize to non-zero values when deploying OceanBase Database and starting the OceanBase cluster:

• If the value of datafile_maxsize is greater than the value of datafile_size or greater than the disk space calculated based on the datafile_disk_percentage ratio, automatic expansion will take effect after the cluster is started.

• If the value of datafile_maxsize is less than the value of datafile_size or less than the disk space calculated based on the datafile_disk_percentage ratio, you need to modify the value of datafile_maxsize to a value greater than datafile_size before enabling automatic expansion of data files.

If only datafile_size (or datafile_disk_percentage) is configured when the cluster is started

If you have only configured datafile_size or datafile_disk_percentage when deploying OceanBase Database and starting the OceanBase cluster, you just need to manually set the values of datafile_maxsize and datafile_next, and ensure that the value of datafile_maxsize is greater than the value of datafile_size or greater than the disk space calculated based on the datafile_disk_percentage ratio.

If datafile_size (or datafile_disk_percentage), datafile_next, and datafile_maxsize are not configured when the cluster is started

If you have not configured datafile_size (or datafile_disk_percentage), datafile_next, and datafile_maxsize when deploying OceanBase Database and starting the OceanBase cluster, you can follow the steps below to enable automatic expansion of disk data files.

The specific operations are as follows:

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

2. Execute the following statement to confirm the pre-allocated space on the disk.

   obclient> SELECT data_disk_allocated/1024/1024/1024 AS datafile_G FROM oceanbase.GV$OB_SERVERS;
   

   Here is an example of the query result.

   +-------------------+
   | datafile_G        |
   +-------------------+
   | 1770.000000000000 |
   +-------------------+
   1 row in set
   

   According to the query result, the pre-allocated space on the disk data_disk_allocated is 1770 GB, approximately 1.7 TB.

   For more information about the fields in the GV$OB_SERVERS view, see GV$OB_SERVERS.

3. Based on the result from the previous step, set the value of datafile_maxsize to a value greater than data_disk_allocated, and set the value of datafile_next to a non-zero value.

   Here is an example of the setting:

   ALTER SYSTEM SET datafile_maxsize='4TB';
   

   ALTER SYSTEM SET datafile_next='1TB';
   

4. After the setting is successful, execute the following statement to confirm whether automatic expansion is enabled.

   SELECT data_disk_allocated/1024/1024/1024 AS datafile_G, data_disk_capacity/1024/1024/1024 AS datafile_max_G FROM oceanbase.GV$OB_SERVERS;
   

   Here is an example of the query result.

   +-------------------+-------------------+
   | datafile_G        | datafile_max_G    |
   +-------------------+-------------------+
   | 1770.000000000000 | 4096.000000000000 |
   +-------------------+-------------------+
   1 row in set
   

   According to the query result, the maximum space for automatic expansion of disk data files data_disk_capacity is set to the value of datafile_maxsize, and the value of data_disk_capacity is greater than the value of data_disk_allocated, indicating that automatic expansion is enabled.

What to do next

After the configuration is successful, the system will automatically expand the data files based on the actual usage of the data files and the values of datafile_next and datafile_maxsize, to minimize the allocation of disk space.

References

• Installation directory structure of an OBServer node

• Deploy OceanBase Database Enterprise Edition

• datafile_size

• datafile_disk_percentage

• datafile_next

• datafile_maxsize

• Modify cluster parameters