This topic describes the release notes for product versions of PolarDB-X.
From April 2024, the release notes include the following changes:
The release notes for product versions are added. For example, features such as columnar storage engines and compatibility with MySQL 8.0 are added in V2.4.0.
The release notes are categorized into release notes for compute nodes, data nodes, Change Data Capture (CDC) nodes, and columnar nodes.
The component version number consists of a three-digit version number and the release date (replacing the original build number).
Version number details
Example | Product version | Component version of compute node | Component version of data node | Engine version |
polardb-2.4.0_5.4.19-20240426_xcluster5.4.19-20240412 | polardb-2.4.0 | 5.4.19-20240426 Note 5.4.19 is the three-digit version number, and 20240426 indicates the release date. | xcluster5.4.19-20240412 Note 5.4.19 is the three-digit version number, and 20240412 indicates the release date. | MySQL 5.7 |
polardb-2.4.0_5.4.19-20240426_xcluster8.4.19-20240430 | xcluster8.4.19-20240430 Note 8.4.19 is the three-digit version number), and 20240430 indicates the release date. | MySQL 8.0 |
Release date: 2024-10-24
Product version: V2.5.0
Category | Description |
New feature and feature optimization | You can execute the DDL statement to create primary tables that have column store indexes defined. You can manually enable or disable this feature. This feature is enabled by default. For more information, see Limits on DDL statements. |
Collecting statistics on databases and tables in parallel is supported. For more information, see Specify a statistics collection policy. | |
When you create a GSI, you can use a globally ordered primary key range to perform concurrent backfill. This improves the performance of GSI building. For more information, see GSIs. | |
The resource limit and scheduling efficiency of logical and physical backfills is improved to boost the stability of DDL execution. | |
The keepalive interval and timeout parameters of the HA system are optimized to shorten the recovery time of the HA system. | |
The performance overheads when metadata locks (MDLs) wait for detection are optimized. For more information, see MDL optimization. | |
Fixed issue | The following issue is fixed: The memory of temporary tables is not released after iterative subqueries are completed. |
The following issue is fixed: When deleting a database coincides with database scaling, a deadlock may occur. | |
The following issue is fixed: If the connection is interrupted when the | |
Release notes | |
Release date: 2024-04-18
Product version: polardb-V2.4.0
Category | Description |
New feature | A cost-effective specification is introduced for read-only columnar nodes. You can use this specification together with different specifications of row store nodes to ensure support for scenarios that involve mixed workloads. |
A global database network (GDN) can be connected, which provides a high-performance, stable, and reliable cross-cloud deployment solution and significantly enhances the performance and reliability of master-replica replication. | |
Column store indexes (CSIs) can be created, and the query optimizer is optimized. This significantly improves the efficiency of complex queries. | |
Fully compatible with MySQL 8.0 protocols and syntax. | |
Feature optimization | The online alteration of column types is optimized to improve stability, and the time required for the alteration is reduced. |
DDL operations, such as adding or removing columns, adding or removing indexes, and modifying columns, can be atomically committed across multiple shards after the operations are pushed down to data nodes. | |
Fixed issue | The following issue is fixed: An error is occasionally reported if a window function includes count. |
The following issue is fixed: Incorrect results may be occasionally returned when you query the timestamp field in an archived table. | |
The following issue is fixed: The calculation result of a row expression is incorrect if the row expression is not pushed down for execution. | |
Release notes | |