LibraDB Engine Core Version Release Notes

Last updated: 2025-12-03 09:58:15

LibraDB Engine Core Version Release Notes

Last updated: 2025-12-03 09:58:15

This document introduces the LibraDB engine kernel version updates of TDSQL-C for MySQL. For detailed product features, see LibraDB Engine Features.
Version Number Rule
The kernel version of the LibraDB engine can be viewed in instance details. The version number is composed of four digits, separated by dots. Different digits represent different version meanings. The details are as follows.
First digit: represents the main version of the product. When the current version of the product is not compatible with the previous version, this version number will be increased by 1.
Second digit: represents the iterative version of the product. When important features are added to the product, this iterative number will be defined based on the current year and month.
Third digit: represents the minor version of the product. When some optimization features or bug fixes are made to the product, this version number will be increased by 1.
Fourth digit: represents the patch version of the product. When it is found that there is a bug urgently to be fixed in the product, a patch upgrade will be performed on this version.
Version Update Description
Update Notes of LibraDB Engine 4.2506.x
Update Notes of LibraDB Engine 3.2503.x
Update Notes of LibraDB Engine 2.2410.x
Update Notes of LibraDB Engine 1.2404.x
Version
Description
4.2506.2.0
Feature updates
Supported the time function.
Supported the yearweek function.
Supported null aware semi join.
Supported the HTAP capability of automatic forwarding to row storage nodes, including error forwarding and missing object forwarding.
The date_add/date_sub function supported composite data types, such as hour_second.
System optimizations
Optimized the action of full scans in primary key pruning scenarios by reducing the data scanning range, thereby improving the query performance.
Optimized the automatic mapping feature, significantly improving its stability.
Bug fixes
Fixed the issue of incorrect results when TopN lazy materialization queries were executed.
Fixed the issue where the database was not specified during HTAP automatic forwarding.
Fixed the issue where the column-format storage secondary index queries of the date type were ineffective.
Fixed the issue of repeated statistics collection on columns with null values only.
Fixed the issue of errors when null strings are converted to JSON data types in comparison operators.
Fixed the issue of errors when the scan is executed on bitmap indexes created by users for the string type columns.
Fixed the issue where select statements involving the dual table were not pushed down to the storage layer for execution in the ETL write-back feature.
4.2506.1.0
The brand new version 4.2506.x supports all features from previous versions, with added support for multiple optimization features. The detailed feature list is as follows:
Feature updates
Supported full-text search. The high-performance full-text search feature was introduced in version 4.2506.x, using a storage framework with deep integration of an ordered dictionary and the Roaring Bitmap technology, significantly improving performance in keyword search and complex text search scenarios.
Supported thread scheduling graph query. The SQL analysis report could generate detailed Processor scheduling graphs for SQL, enabling easier analysis and location of SQL performance bottlenecks.
Supported viewing SQL CPU time, execution time of the Join module, and peak memory usage of Join, Agg, and Sort operators in the SQL analysis report.
Supported the Work steal feature, addressing the uneven data stream issue after index and predicate filtering.
Comprehensively enhanced statistical information. A brand-new histogram collection framework was supported, and the composite histogram collection was added. Index statistics and the locking/unlocking of statistical information were also supported.
Supported the brand-new partition pruning capability to automatically remove irrelevant partitions during the query, thereby improving query performance.
Supported end-to-end Optimizer Trace, showing details and the decision-making process at each optimization stage to help developers understand SQL optimization in depth.
Supported the display of the Explain Analyze plan to show the execution plan of an SQL query and actually run the query, outputting more detailed execution plans and time consumption of execution steps.
Supported the Recursive CTE syntax, which can be used for hierarchical and tree-structured data queries, enhancing query efficiency for such data structures.
System optimizations
Optimized index Filter pruning logic, refined pruning capacity by Filter semantics, and expanded the index-supported scenarios.
Optimized Rowset pruning logic to address the issue of extremely high threading caused by high-frequency memory applications under high-concurrency scenarios.
Optimized the display information in SQL analysis reports and increased the information display for the lazy materialization operator.
Further enhanced the performance, with a 17.5% improvement compared to version 3.2503.x. For details, see TPC-H Performance Results.
Version
Description
3.2503.7.0
Feature updates
Supported the Truncate function.
Supported the HOUR_SECOND type in the date_add and date_sub functions.
Supported the right outer join query capability.
Supported the ability to route queries for tables not loaded into the analysis engine to read-write nodes for execution.
Bug fixes
Fixed the issue of the query error when tables were synchronized to columnar storage in case-sensitive scenarios.
Fixed the issue of Bitmap index creation errors.
3.2503.6.0
Feature updates
Supported the Timediff function.
System optimizations
Optimized the outer Join elimination capability in multi-table Join scenarios.
Optimized the method to obtain constant values when adding partitions to a partition table, avoiding the impact from session variables.
Bug fixes
Fixed the issue of excessive resource consumption in metadata management when there are too many tables.
Fixed the issue of synchronization exceptions caused by reloading objects in case-sensitive scenarios.
3.2503.5.0
Feature updates
Supported the traffic throttling capability during backup.
Supported security reinforcement for the backup process.
Supported the UUID function.
Supported loading partitioned tables that include constants or function expressions into the analysis engine using the LESS THAN syntax.
Supported the capability of generating SQL analysis reports.
Supported the capability of forwarding SQL that is unsuitable for column-format storage execution to the specified row storage instances for execution.
System optimizations
Optimized the slow log recording format of the analysis engine and added the number of rows and the size of the result set.
Optimized and unified the PROFILE naming style for SCAN and lazy materialization (COLUMN_READ) operators.
Bug fixes
Fixed the issue that the Runtime filter failed to be pushed down to Union All.
Fixed the issue of handling errors when certain table mappings are updated in all-in-one scenarios.
Fixed the issue of abnormal synchronization statuses of objects in all-in-one scenarios.
Fixed the issue of abnormal synchronization of DDL changes for partition tables in all-in-one scenarios.
Fixed the query error when certain fields in the composite primary key include NULL values.
Fixed the issue that the COMMENT is not synchronized when the CREATE LIKE statement is synchronized.
3.2503.4.0
This version merges versions 3.2503.2.0 and 3.2503.3.0. The change information is as follows:
Feature updates
Supported synchronizing partition tables that used UNIX_TIMESTAMP and TO_DAYS functions as partition functions to the analysis engine.
Supported the implicit conversion capability when the GREATEST function compared data of the DATETIME and character types.
Supported the equivalent predicate derivation capability in Apply scenarios.
Supported DDL scenarios of adding a column while specifying a default value of the time type (value of Datetime).
Supported data synchronization for DDL statements involving multiple operations, such as column deletion and column modification.
Supported synchronizing tables with nullable columns in the composite primary key to the analysis engine.
Supported customizing sorting keys for data synchronization to columnar storage tables to significantly improve the query efficiency of SQL statements based on sorting columns.
Supported the FIELD function.
Bug fixes
Fixed the issue of collecting histogram statistics for unexpired columns during automatic statistics collection.
Fixed the issue that the SQL statements were truncated when SLOW_LOG was queried.
Fixed the issue of pushing down complex predicates to avoid pushing down irrelevant predicates to On-Condition.
Fixed the issue that the table name in metadata failed to be modified accordingly after the name was modified.
Fixed the issue of error reporting caused by lazy materialization when certain SQL statements were executed.
Fixed the issue that columnar storage indexes failed to take effect in Join scenarios.
Fixed the issue that the Bloom filter index failed to take effect when the IN-conditions list contained too many values.
System optimizations
Optimized the logic of displaying warning prompts during SQL queries in data merge scenarios.
Optimized the logic of automatically converting table definition names into lowercase names in data merge scenarios.
Optimized the efficiency of writing data to the disk during full data loading.
Optimized the determination mechanism in automatic forwarding scenarios to improve determination accuracy.
Optimized the process of executing DDL operations to remove columns. During DDL execution, the system verifies whether NULL values exist to avoid storage errors.
Optimized the index addition logic to allow only one index of the same type for the same column, thus avoiding repeated index creation.
3.2503.1.0
Feature updates
Supported multi-source data merge and database/table merge capabilities to aggregate data scattered across different database instances via LibraDB for analysis and query.
Supported cost-based decision-making to forward SQL requests to specified row storage nodes for execution. This feature is provided as a testing feature in the current version and is disabled by default.
Supported the backup and recovery capability for quick backup and recovery of columnar storage data.
Supported the static data archiving capability to store certain static data in Cloud Object Storage (COS).
Supported fixed-length types compatible with MySQL Char types and the space addition/removal actions for fixed-length data types.
Supported the logical CPU-based queuing mechanism, and introduced query-based mechanisms of cost-based ranking for logical plans and point lookup detection. The query efficiency was significantly improved in certain mixed workload scenarios.
Supported the Rollup syntax. The Rollup syntax is mainly used for grouping and querying. It is an extension of Group By and can generate multi-level aggregated results, including results of subgroups and total results.
Supported NestedLoopJoin. This operator can be used in scenarios with non-equivalent associations of primary keys or indexes. It can significantly improve performance in certain primary key join scenarios. It also provides the computing capability to dissociate relevant subqueries in Join On expressions of the optimizer.
System optimizations
Optimized the global thread pool capability. Thread usage can be adjusted according to the actual business situation to avoid creating too many threads and improve system stability and resource utilization.
Optimized grace hash join to use the state machine to optimize the thread waiting logic under execution, thus fixing thread bloat issues.
Optimized the memory module to collect the current memory usage by features and log memory usage in more detail.
Version
Description
2.2410.14.0
Feature updates
Supported the time function.
Supported the yearweek function.
Supported null aware semi join.
Bug fixes
Fixed the issue of incorrect results when TopN lazy materialization queries were executed.
Fixed the issue where the database was not specified during HTAP automatic forwarding.
Fixed the issue where the column-format storage secondary index queries of the date type were ineffective.
Fixed the issue of repeated statistics collection on columns with null values only.
2.2410.13.0
Feature updates
Supported the Truncate function.
Supported the Timestamp function.
Supported the HOUR_SECOND type in the date_add and date_sub functions.
Supported the right outer join query capability and supported the curdate function as the default value of the column.
Supported the ability to route queries for tables not loaded into the analysis engine to read-write nodes for execution.
System optimizations
Optimized the performance of Order By queries.
Bug fixes
Fixed the issue of the query error when tables were synchronized to columnar storage in case-sensitive scenarios.
Fixed the issue of Bitmap index creation errors.
2.2410.12.0
Feature updates
Supported the backup and recovery feature of the analysis engine and traffic throttling during backup.
Supported security reinforcement for the backup process.
Supported the UUID function.
Supported loading partitioned tables that include constants or function expressions into the analysis engine using the LESS THAN syntax.
Supported the capability of generating SQL analysis reports.
System optimizations
Optimized the slow log recording format of the analysis engine and added the number of rows and the size of the result set.
Optimized and unified the PROFILE naming style for SCAN and lazy materialization (COLUMN_READ) operators.
Bug fixes
Fixed the issue of abnormal synchronization statuses of objects in all-in-one scenarios.
Fixed the query error when certain fields in the composite primary key include NULL values.
Fixed the issue that the COMMENT is not synchronized when the CREATE LIKE statement is synchronized.
2.2410.11.1
Feature updates
Supported scenarios with nested functions in the partition table definition statement less than.
Supported streaming write-back for the ETL write-back feature to avoid write timeout caused by too much data written at one time.
Supported adjusting the ordering key of column-format storage tables by adding COMMENT in the table definition statement. For details, see Customizing Column-Format Storage Ordering Keys.
Supported Nullable columns as part of the ordering key.
System optimizations
Skipped permission verification for DML statements during the execution of the PREPARE statement to avoid "unable to find" errors in the analysis engine when the PREPARE statement containing DML statements is being executed in the database agent process.
Optimized the display of detailed information for the lazy materialization operator in SQL analysis reports.
Bug fixes
Fixed system errors caused by comparison operations on JSON data.
Fixed errors reported by the system when limit 1 is used in subqueries.
Fixed errors reported when the Runtime filter is pushed down to Union All.
2.2410.10.1
This version merges the changes in version 2.2410.9.0. The change information is as follows:
Feature updates
Supported synchronizing partition tables that used UNIX_TIMESTAMP and TO_DAYS functions as partition functions to the analysis engine.
Supported the implicit conversion capability when the GREATEST function compared data of the DATETIME and character types.
Supported the equivalent predicate derivation capability in Apply scenarios.
Supported synchronizing tables with Generated Invisible Primary Key (GIPK) to the analysis engine.
Supported DDL scenarios of adding a column while specifying a default value of the time type (value of Datetime).
Supported data synchronization for DDL statements involving multiple operations, such as column deletion and column modification.
Bug fixes
Fixed the issue of collecting histogram statistics for unexpired columns during automatic statistics collection.
Fixed the issue that the SQL statements were truncated when SLOW_LOG was queried.
Fixed the issue that columnar storage indexes failed to take effect in Join scenarios.
Fixed the issue of pushing down complex predicates to avoid pushing down irrelevant predicates to On-Condition.
Fixed the issue that the Bloom filter index failed to take effect when the IN-conditions list contained too many values.
Fixed the issue of database/table reloading during synchronized object modification.
Fixed the issue of inconsistent time zone with the primary instance during data synchronization.
Fixed the issue that the table name in metadata failed to be modified accordingly after the name was modified.
System optimizations
Optimized data skew of lazy materialization during table queries to improve execution performance.
Optimized the index addition logic to allow only one index of the same type for the same column, thus avoiding repeated index creation.
2.2410.8.0
System optimizations
Optimized the default value of the concurrency of writing data to the disk, greatly enhancing the execution efficiency of full data loading.
Optimized the logic for generating query plans, resulting in performance improvement in certain query scenarios.
Bug fixes
Resolved an issue where special characters, such as "artistic characters", in the data could not be loaded as columnar storage.
2.2410.7.0
Feature updates
Supported row and column synchronization in scenarios involving PT changes, Gh-ost changes, and Alibaba Cloud DMS lock-free changes for "read-write nodes".
Supported the predicate derivation capability in outer join scenarios.
Opened configuration for the block_cache_capacity_mb parameter.
Supported scenarios where ON expressions contain subqueries.
Enabled synchronization of data tables with column names containing % to the analysis engine.
System optimizations
Optimized the execution performance of Runtime Filter.
Optimized the partition pruning performance for partition tables.
Optimized thread usage in high-concurrency execution scenarios.
Bug fixes
Fixed the error encountered when Union operations are performed for Tinyint and Int data.
Fixed the data error that might occur when data incremental synchronization was paused.
Fixed the issue that synchronization cannot be performed when Create View contains Union.
Fixed the execution error encountered during the execution of PREPARE statements.
Fixed the issue of query exceptions due to expression pushdown.
Fixed the query errors encountered when Bitmap indexes are executed.
2.2410.6.0
Feature updates
Supported setting default values for columns when adding a column.
System optimizations
Optimized slow SQL information displayed in the system table SLOW_LOG to avoid too many invalid SQL executions that may cause interference to users.
Optimized the timing selection policy for executing Compact operations on backend storage.
Optimized the thread volume expansion issue during concurrent execution.
2.2410.5.0
Feature updates
Supported aggregate function calculation on element values fetched from JSON data types.
Supported status prompt when data full loading is complete.
Supported Union operations on Decimal type fields with different lengths from multiple Select statement results.
System optimizations
Optimized the logic of partial collection metrics during data loading, improving the accuracy of monitoring metrics.
2.2410.4.1
Bug fixes
Fixed table structure inconsistency that may occur during the Add Column operation.
System optimizations
Improved the connection release efficiency when accessing the read-only analysis engine through Database Proxy.
2.2410.4.0
Feature updates
Supported the brand-new database proxy version 1.4.4.
Supported users to create a columnar storage secondary index.
Bug fixes
Fixed the issue where the Sort/TopN operator was unable to generate a multi-table Colocate Join.
Fixed the error and alarm information when loading part of the DDL.
Fixed the issue where some operators cannot be profiled.
Fixed the issue where too many logs were performed during the data loading process, occupying too much storage space.
Fixed the issue where the number of affected rows returned by the MySQL client and the Records entry does not correspond after the ETL writeback feature is executed. 
2.2410.3.0
Bug fixes
Fixed the issue where Join cannot be performed between Uint64 and Int64 field types.
Fixed the issue where data loading was interrupted when encountering the Geometry data type during data loading.
Fixed the issue where a parsing error occurred when creating an index on a null column.
Fixed a logic issue where the table could not be reconstructed after deleting the primary key and then immediately creating a primary key during incremental data loading.
System optimizations
Optimized the processing policy of tables without primary key when loading data, making the loading of tables without primary key more stable.
2.2410.2.0
Feature updates
Added support for the Date_Add/Find_In_Set/Weekday functions.
Bug fixes
Fixed the issue where excessive memory usage was caused by using Cross Join when adding a subquery in an OR expression.
Fixed the issue where an error is occasionally reported when running the command if there is a constant column in SQL statements.
Fixed the issue where an error is occasionally reported when running the command under an IF Expression.
Fixed the issue where an execution error is reported when processing the column length in some expressions.
2.2410.1.0
Feature updates
Storage features
Added support for columnar storage secondary index. The supported index types include the Zonemap index, Bloom Filter index, and Bitmap index.
Added support for concurrent FlushDMS operations to enhance data write efficiency, particularly improving update efficiency when managing large volumes of data.
Added support for the JSON type and various JSON functions.
Added support for loading range/list-partitioned tables into the LibraDB engine and loading partition DDL statements synchronously. For detailed syntax and function support related to partition operations (such as adding or dropping partitions or subpartitions), see Data Loading Restrictions.
Added support for loading tables without primary keys in a columnar storage format, as well as for scenarios involving modifications to table primary keys. For details, see Data Loading Restrictions.
Compute features
Added support for Merge join, improving query performance in primary key join scenarios.
Added support for stream aggregation, improving GROUP BY performance in ordered data scenarios.
Added support for the automatic statistics collection capability, removing the need for users to manually collect statistics regularly.
Added support for the random sampling feature, enhancing the efficiency of statistics collection and reducing the resource consumption associated with statistics.
Added support for adaptive GROUP BY operators.
Added support for streaming CTEs. For details, see CTE Syntax Use Instructions.
Added support for pushing Table Dual down to the storage layer for execution.
Added support for pushing the UNION ALL operator down to the storage layer for execution.
Added support for the Win Magic rewrite of the optimizer, which decouples correlated subqueries by rewriting them as window functions to improve query efficiency.
Added support for 14 MySQL-compatible functions. For details, see Description of Supported String Functions.
Added support for string-type input values for the Hour, Minute, Second, and Microsecond functions.
Added support for loading tables with decimal-type primary keys (up to a maximum length of 256) into columnar storage.
Added support for NULL-aware anti-joins. When executing set operations, these joins intelligently assess whether a set is empty or contains null values, optimizing the execution efficiency of operations such as NOT IN and <>ALL. This enhancement significantly improves the performance of associated SQL queries.
System optimizations
Optimized the error messages returned for some execution errors.
Optimized the performance of the grace hash join by employing a two-phase bucket acceleration mechanism, reducing the frequency of bucket scale-out.
Added support for right anti-joins/right semi-joins in the result set splitting feature, significantly alleviating long wait times for data reception caused by extremely large join result sets.
Added support for aggregation in the result set splitting feature, alleviating long wait times for data reception caused by extremely large aggregation output result sets.
Optimized the EXPLAIN feature by enabling the display of the table name and alias corresponding to the TableScan operator in the detailed information of the execution plan.
Optimized the EXPLAIN feature by enabling the display of replica information for the referenced table in the execution plan.
Added support for scenarios where the Unique Key column contains null values.
Optimized the metadata storage mechanism, significantly reducing excessive resource consumption associated with metadata.
Optimized the execution method of the Apply operator, improving query performance in correlated subquery scenarios.
Version
Description
1.2404.24.0
Feature updates
Supported setting default values for columns when adding a column.
System optimizations
Optimized slow SQL information displayed in the system table SLOW_LOG to avoid too many invalid SQL executions that may cause interference to users.
Optimized the timing selection policy for executing Compact operations on backend storage.
Optimized the thread volume expansion issue during concurrent execution.
1.2404.23.0
Feature updates
Supported status prompt when data full loading is complete.
System optimizations
Optimized the logic of partial collection metrics during data loading, improving the accuracy of monitoring metrics.
1.2404.22.1
Bug fixes
Fixed table structure inconsistency that may occur during the Add Column operation.
System optimizations
Improved the connection release efficiency when accessing the read-only analysis engine through Database Proxy.
1.2404.22.0
This version is an issue-fixed and optimized version, primarily addressing defects in versions before 1.2404.21.0 (inclusive).
Bug fixes
Fixed the data latency issue caused by synchronizing DDL in some scenarios.
Fixed the issue where an error is reported when constructing a Datetime type for a correlated subquery column.
Fixed the issue where an error is reported when the parameter is an array in the scenario of an IN expression.
System optimizations
Improved the performance when querying with a primary key predicate under a large data volume.
Improved the truncation length of slow SQL query views when encountering relatively long SQL.
1.2404.21.0
This release is a bugfixing and optimization update that primarily addresses defects identified in version 1.2404.20.0 and earlier.
Feature updates
Added support for DDL operations related to partitioned tables in the current version.
Bug fixes
Resolved the issue that data deletion failed when the primary key contained a timestamp type.
Resolved the data synchronization error associated with SET type data.
Resolved several errors in the table renaming scenario.
1.2404.20.0
This release is a bugfixing and optimization update that primarily addresses defects identified in version 1.2404.19.1 and earlier.
Bug fixes
Resolved the issue that caused data-loading objects to be inaccessible when the read-write instance was configured as case-sensitive.
Resolved the issue that some existing instances could not support database proxies.
Resolved the issue that Lazy Materialization statistics were displayed incorrectly.
Resolved the issue that DMC was unable to retrieve user information for the read-only analysis engine.
Optimized the default values for some kernel parameter settings.
1.2404.19.1
This release is a bugfix update that primarily addresses defects identified in versions earlier than 1.2404.19.0.
Bug fixes
Resolved the issue of inconsistent view definers when loading views into the read-only analysis engine.
Resolved the issue that user synchronization failed for users with empty passwords in the mysql.user system table.
Resolved the execution error encountered when executing DDL statements during SQL operations.
Fixed the issue where the "ETL Writeback Acceleration" feature encounters a writing exception when there is a constant column.
1.2404.19.0
Feature updates
Added support for accessing the read-only analysis engine through the database proxy when using the HINT syntax.
Added support for accessing the read-only analysis engine through DMC.
System optimizations
Optimized the memory management methods for the SORT and AGGREGATION operators, enhancing memory usage efficiency. Following the optimization, memory used by the SORT and AGGREGATION operators can be released immediately upon completion of their execution, without waiting for the entire query execution to end.
Optimized the memory usage method for join result output, reducing memory consumption when the JOIN operator produces a relatively large result set.
Optimized the allocation policy for Lazy Materialization in primary key scenarios.
Bug fixes
Resolved the issue that PREPARE STATEMENT defaulted to using Plan Cache, resulting in multiple similar queries inadvertently matching a previously cached plan. This could lead to query errors when the types of constant parameters were inconsistent.
Resolved the issue of query errors that occurred in some scenarios when output columns of VARCHAR type were used in a UNION ALL context.
Resolved the issue that metadata retrieval errors were caused when DMC connected to the columnar storage engine due to insufficient permissions.
Fixed a potential memory leak issue in the scenario where joins are created frequently.
Resolved the issue that temporary tables were prevented from being queried in the read-only analysis engine in scenarios involving changes to partitioned tables.
Resolved the issue that the Lazy Materialization operator was not displayed in the execution plan under some scenarios.
Resolved the issue that the read-only analysis engine was unable to delete fields from the join index.
Resolved the out-of-bounds error in runtime filter assignment for multi-table join scenarios.
Resolved the issue that the runtime filter occasionally returned empty data.
Resolved the error caused by a mismatch between nullable and non-nullable types in UNION ALL scenarios.
1.2404.17.0
Feature updates
Added support for the FROM_UNIXTIME and UNIX_TIMESTAMP functions.
System optimizations
Optimized the permissions required to query system tables when database client tools, such as DBearer, access the read-only analysis engine.
Optimized error messages for some unsupported functions.
Bug fixes
Resolved the issue that the read-only analysis engine incorrectly loaded Timestamp field values when the explicit_defaults_for_timestamp parameter differed across instances.
1.2404.16.0
Feature updates
Added support for using the read-only analysis engine to accelerate the execution of query statements in INSERT…SELECT…, enabling faster writes to read-write instances. For details, see ETL Writeback Acceleration.
Added support for querying and synchronizing data types, including BIT, ENUM, and SET.
System optimizations
Improved the naming format of some system tables to make them easier to understand, such as the SLOW_LOG table.
Adjusted the supported value size for regular columns to 16 MB, up from the previous limit of 5 MB.
Optimized the logic of the StorageTableReader operator for retrieving column names when the query plan is empty. 
Bug fixes
Resolved the issue that, in subquery scenarios, the Apply operator logic was still applied when the primary query result was empty, leading to a query error.
Resolved the issue that the Apply operator caused a crash during multi-threaded competition.
1.2404.15.2
This release is a bugfixing update that primarily addresses defects identified in versions earlier than 1.2404.15.1.
Bug fixes
Resolved the issue that the instance crashed during data queries when high-concurrency UPDATE operations were performed on the read-write instance after it was loaded as a column-stored table.
1.2404.15.1
System optimizations
Optimized the default settings for database log retention. Specifically, if an instance's disk size is 500 GB or less, the default maximum log retention is set to 5 GB; for instances with a disk size over 500 GB, it is set to 10 GB.
Enabled Lazy Materialization in the read-only analysis engine by default. For details on its capabilities and tuning options, see Lazy Materialization.
Added support for loading tables whose names begin with an underscore (_) into the read-only analysis engine.
Improved the restart efficiency of the read-only analysis engine.
Bug fixes
Resolved the issue that some versions of Navicat displayed an insufficient permissions error when connecting to the read-only analysis engine.
Resolved the issue that the read-only analysis engine failed to load data when DDL statements were executed consecutively.
Resolved the syntax compatibility issue related to DEFAULT (0) in the CREATE TABLE statement.
1.2404.10.0
Feature updates
Added support for the DATE_SUB function.
Added support for users to manually collect table statistics. For details, see Collecting Statistics.
Added support for synchronous loading of ENUM data in special scenarios.
Note:
Although ENUM data can be loaded into the read-only analysis engine in special scenarios, querying data of the ENUM field type is still not supported.
System optimizations
Adjusted the kernel to manage the number of versions retained for expired data in storage, preventing excessive disk space usage caused by retaining an excessive number of data snapshots.
Optimized kernel resource allocation during data clearing, preventing degradation of user SQL execution performance resulting from excessive consumption of system resources.
Adjusted the logging level for kernel information to prevent excessive log output from consuming too much disk space.
Adjusted the maximum value of the storage field. The previously supported default limit of 1 MB has been increased to 5 MB.
Optimized the error message for data querying when the data is not fully loaded into the read-only analysis engine. It now clearly indicates that you need to wait until all data is fully loaded before performing a query.
Optimized the error message for unsupported data types and functions during execution. You can now clearly identify the unsupported items in the error message.
Optimized the internal trigger frequency and selection logic for background tasks, ensuring they do not impact the execution performance of users' online SQL queries.
Improved the version number information displayed when accessing the read-only analysis engine instance through the MySQL client to prevent potential misunderstandings.
1.2404.7
Feature updates
Added support for real-time loading of hyperscale data into the read-only analysis engine.
Added support for ultra-high performance in complex data queries.
Note:
This is the first launched LibraDB engine kernel version for TDSQL-C for MySQL.
﻿

Was this page helpful?

You can also Contact Sales or Submit a Ticket for help.

Yes

Feedback

Version	Description
4.2506.2.0	Feature updates Supported the time function. Supported the yearweek function. Supported null aware semi join. Supported the HTAP capability of automatic forwarding to row storage nodes, including error forwarding and missing object forwarding. The date_add/date_sub function supported composite data types, such as hour_second. System optimizations Optimized the action of full scans in primary key pruning scenarios by reducing the data scanning range, thereby improving the query performance. Optimized the automatic mapping feature, significantly improving its stability. Bug fixes Fixed the issue of incorrect results when TopN lazy materialization queries were executed. Fixed the issue where the database was not specified during HTAP automatic forwarding. Fixed the issue where the column-format storage secondary index queries of the date type were ineffective. Fixed the issue of repeated statistics collection on columns with null values only. Fixed the issue of errors when null strings are converted to JSON data types in comparison operators. Fixed the issue of errors when the scan is executed on bitmap indexes created by users for the string type columns. Fixed the issue where select statements involving the dual table were not pushed down to the storage layer for execution in the ETL write-back feature.
4.2506.1.0	The brand new version 4.2506.x supports all features from previous versions, with added support for multiple optimization features. The detailed feature list is as follows: Feature updates Supported full-text search. The high-performance full-text search feature was introduced in version 4.2506.x, using a storage framework with deep integration of an ordered dictionary and the Roaring Bitmap technology, significantly improving performance in keyword search and complex text search scenarios. Supported thread scheduling graph query. The SQL analysis report could generate detailed Processor scheduling graphs for SQL, enabling easier analysis and location of SQL performance bottlenecks. Supported viewing SQL CPU time, execution time of the Join module, and peak memory usage of Join, Agg, and Sort operators in the SQL analysis report. Supported the Work steal feature, addressing the uneven data stream issue after index and predicate filtering. Comprehensively enhanced statistical information. A brand-new histogram collection framework was supported, and the composite histogram collection was added. Index statistics and the locking/unlocking of statistical information were also supported. Supported the brand-new partition pruning capability to automatically remove irrelevant partitions during the query, thereby improving query performance. Supported end-to-end Optimizer Trace, showing details and the decision-making process at each optimization stage to help developers understand SQL optimization in depth. Supported the display of the Explain Analyze plan to show the execution plan of an SQL query and actually run the query, outputting more detailed execution plans and time consumption of execution steps. Supported the Recursive CTE syntax, which can be used for hierarchical and tree-structured data queries, enhancing query efficiency for such data structures. System optimizations Optimized index Filter pruning logic, refined pruning capacity by Filter semantics, and expanded the index-supported scenarios. Optimized Rowset pruning logic to address the issue of extremely high threading caused by high-frequency memory applications under high-concurrency scenarios. Optimized the display information in SQL analysis reports and increased the information display for the lazy materialization operator. Further enhanced the performance, with a 17.5% improvement compared to version 3.2503.x. For details, see TPC-H Performance Results.

Version	Description
3.2503.7.0	Feature updates Supported the Truncate function. Supported the HOUR_SECOND type in the date_add and date_sub functions. Supported the right outer join query capability. Supported the ability to route queries for tables not loaded into the analysis engine to read-write nodes for execution. Bug fixes Fixed the issue of the query error when tables were synchronized to columnar storage in case-sensitive scenarios. Fixed the issue of Bitmap index creation errors.
3.2503.6.0	Feature updates Supported the Timediff function. System optimizations Optimized the outer Join elimination capability in multi-table Join scenarios. Optimized the method to obtain constant values when adding partitions to a partition table, avoiding the impact from session variables. Bug fixes Fixed the issue of excessive resource consumption in metadata management when there are too many tables. Fixed the issue of synchronization exceptions caused by reloading objects in case-sensitive scenarios.
3.2503.5.0	Feature updates Supported the traffic throttling capability during backup. Supported security reinforcement for the backup process. Supported the UUID function. Supported loading partitioned tables that include constants or function expressions into the analysis engine using the LESS THAN syntax. Supported the capability of generating SQL analysis reports. Supported the capability of forwarding SQL that is unsuitable for column-format storage execution to the specified row storage instances for execution. System optimizations Optimized the slow log recording format of the analysis engine and added the number of rows and the size of the result set. Optimized and unified the PROFILE naming style for SCAN and lazy materialization (COLUMN_READ) operators. Bug fixes Fixed the issue that the Runtime filter failed to be pushed down to Union All. Fixed the issue of handling errors when certain table mappings are updated in all-in-one scenarios. Fixed the issue of abnormal synchronization statuses of objects in all-in-one scenarios. Fixed the issue of abnormal synchronization of DDL changes for partition tables in all-in-one scenarios. Fixed the query error when certain fields in the composite primary key include NULL values. Fixed the issue that the COMMENT is not synchronized when the CREATE LIKE statement is synchronized.
3.2503.4.0	This version merges versions 3.2503.2.0 and 3.2503.3.0. The change information is as follows: Feature updates Supported synchronizing partition tables that used UNIX_TIMESTAMP and TO_DAYS functions as partition functions to the analysis engine. Supported the implicit conversion capability when the GREATEST function compared data of the DATETIME and character types. Supported the equivalent predicate derivation capability in Apply scenarios. Supported DDL scenarios of adding a column while specifying a default value of the time type (value of Datetime). Supported data synchronization for DDL statements involving multiple operations, such as column deletion and column modification. Supported synchronizing tables with nullable columns in the composite primary key to the analysis engine. Supported customizing sorting keys for data synchronization to columnar storage tables to significantly improve the query efficiency of SQL statements based on sorting columns. Supported the FIELD function. Bug fixes Fixed the issue of collecting histogram statistics for unexpired columns during automatic statistics collection. Fixed the issue that the SQL statements were truncated when SLOW_LOG was queried. Fixed the issue of pushing down complex predicates to avoid pushing down irrelevant predicates to On-Condition. Fixed the issue that the table name in metadata failed to be modified accordingly after the name was modified. Fixed the issue of error reporting caused by lazy materialization when certain SQL statements were executed. Fixed the issue that columnar storage indexes failed to take effect in Join scenarios. Fixed the issue that the Bloom filter index failed to take effect when the IN-conditions list contained too many values. System optimizations Optimized the logic of displaying warning prompts during SQL queries in data merge scenarios. Optimized the logic of automatically converting table definition names into lowercase names in data merge scenarios. Optimized the efficiency of writing data to the disk during full data loading. Optimized the determination mechanism in automatic forwarding scenarios to improve determination accuracy. Optimized the process of executing DDL operations to remove columns. During DDL execution, the system verifies whether NULL values exist to avoid storage errors. Optimized the index addition logic to allow only one index of the same type for the same column, thus avoiding repeated index creation.
3.2503.1.0	Feature updates Supported multi-source data merge and database/table merge capabilities to aggregate data scattered across different database instances via LibraDB for analysis and query. Supported cost-based decision-making to forward SQL requests to specified row storage nodes for execution. This feature is provided as a testing feature in the current version and is disabled by default. Supported the backup and recovery capability for quick backup and recovery of columnar storage data. Supported the static data archiving capability to store certain static data in Cloud Object Storage (COS). Supported fixed-length types compatible with MySQL Char types and the space addition/removal actions for fixed-length data types. Supported the logical CPU-based queuing mechanism, and introduced query-based mechanisms of cost-based ranking for logical plans and point lookup detection. The query efficiency was significantly improved in certain mixed workload scenarios. Supported the Rollup syntax. The Rollup syntax is mainly used for grouping and querying. It is an extension of Group By and can generate multi-level aggregated results, including results of subgroups and total results. Supported NestedLoopJoin. This operator can be used in scenarios with non-equivalent associations of primary keys or indexes. It can significantly improve performance in certain primary key join scenarios. It also provides the computing capability to dissociate relevant subqueries in Join On expressions of the optimizer. System optimizations Optimized the global thread pool capability. Thread usage can be adjusted according to the actual business situation to avoid creating too many threads and improve system stability and resource utilization. Optimized grace hash join to use the state machine to optimize the thread waiting logic under execution, thus fixing thread bloat issues. Optimized the memory module to collect the current memory usage by features and log memory usage in more detail.

Version	Description
2.2410.14.0	Feature updates Supported the time function. Supported the yearweek function. Supported null aware semi join. Bug fixes Fixed the issue of incorrect results when TopN lazy materialization queries were executed. Fixed the issue where the database was not specified during HTAP automatic forwarding. Fixed the issue where the column-format storage secondary index queries of the date type were ineffective. Fixed the issue of repeated statistics collection on columns with null values only.
2.2410.13.0	Feature updates Supported the Truncate function. Supported the Timestamp function. Supported the HOUR_SECOND type in the date_add and date_sub functions. Supported the right outer join query capability and supported the curdate function as the default value of the column. Supported the ability to route queries for tables not loaded into the analysis engine to read-write nodes for execution. System optimizations Optimized the performance of Order By queries. Bug fixes Fixed the issue of the query error when tables were synchronized to columnar storage in case-sensitive scenarios. Fixed the issue of Bitmap index creation errors.
2.2410.12.0	Feature updates Supported the backup and recovery feature of the analysis engine and traffic throttling during backup. Supported security reinforcement for the backup process. Supported the UUID function. Supported loading partitioned tables that include constants or function expressions into the analysis engine using the LESS THAN syntax. Supported the capability of generating SQL analysis reports. System optimizations Optimized the slow log recording format of the analysis engine and added the number of rows and the size of the result set. Optimized and unified the PROFILE naming style for SCAN and lazy materialization (COLUMN_READ) operators. Bug fixes Fixed the issue of abnormal synchronization statuses of objects in all-in-one scenarios. Fixed the query error when certain fields in the composite primary key include NULL values. Fixed the issue that the COMMENT is not synchronized when the CREATE LIKE statement is synchronized.
2.2410.11.1	Feature updates Supported scenarios with nested functions in the partition table definition statement less than. Supported streaming write-back for the ETL write-back feature to avoid write timeout caused by too much data written at one time. Supported adjusting the ordering key of column-format storage tables by adding COMMENT in the table definition statement. For details, see Customizing Column-Format Storage Ordering Keys. Supported Nullable columns as part of the ordering key. System optimizations Skipped permission verification for DML statements during the execution of the PREPARE statement to avoid "unable to find" errors in the analysis engine when the PREPARE statement containing DML statements is being executed in the database agent process. Optimized the display of detailed information for the lazy materialization operator in SQL analysis reports. Bug fixes Fixed system errors caused by comparison operations on JSON data. Fixed errors reported by the system when limit 1 is used in subqueries. Fixed errors reported when the Runtime filter is pushed down to Union All.
2.2410.10.1	This version merges the changes in version 2.2410.9.0. The change information is as follows: Feature updates Supported synchronizing partition tables that used UNIX_TIMESTAMP and TO_DAYS functions as partition functions to the analysis engine. Supported the implicit conversion capability when the GREATEST function compared data of the DATETIME and character types. Supported the equivalent predicate derivation capability in Apply scenarios. Supported synchronizing tables with Generated Invisible Primary Key (GIPK) to the analysis engine. Supported DDL scenarios of adding a column while specifying a default value of the time type (value of Datetime). Supported data synchronization for DDL statements involving multiple operations, such as column deletion and column modification. Bug fixes Fixed the issue of collecting histogram statistics for unexpired columns during automatic statistics collection. Fixed the issue that the SQL statements were truncated when SLOW_LOG was queried. Fixed the issue that columnar storage indexes failed to take effect in Join scenarios. Fixed the issue of pushing down complex predicates to avoid pushing down irrelevant predicates to On-Condition. Fixed the issue that the Bloom filter index failed to take effect when the IN-conditions list contained too many values. Fixed the issue of database/table reloading during synchronized object modification. Fixed the issue of inconsistent time zone with the primary instance during data synchronization. Fixed the issue that the table name in metadata failed to be modified accordingly after the name was modified. System optimizations Optimized data skew of lazy materialization during table queries to improve execution performance. Optimized the index addition logic to allow only one index of the same type for the same column, thus avoiding repeated index creation.
2.2410.8.0	System optimizations Optimized the default value of the concurrency of writing data to the disk, greatly enhancing the execution efficiency of full data loading. Optimized the logic for generating query plans, resulting in performance improvement in certain query scenarios. Bug fixes Resolved an issue where special characters, such as "artistic characters", in the data could not be loaded as columnar storage.
2.2410.7.0	Feature updates Supported row and column synchronization in scenarios involving PT changes, Gh-ost changes, and Alibaba Cloud DMS lock-free changes for "read-write nodes". Supported the predicate derivation capability in outer join scenarios. Opened configuration for the block_cache_capacity_mb parameter. Supported scenarios where ON expressions contain subqueries. Enabled synchronization of data tables with column names containing % to the analysis engine. System optimizations Optimized the execution performance of Runtime Filter. Optimized the partition pruning performance for partition tables. Optimized thread usage in high-concurrency execution scenarios. Bug fixes Fixed the error encountered when Union operations are performed for Tinyint and Int data. Fixed the data error that might occur when data incremental synchronization was paused. Fixed the issue that synchronization cannot be performed when Create View contains Union. Fixed the execution error encountered during the execution of PREPARE statements. Fixed the issue of query exceptions due to expression pushdown. Fixed the query errors encountered when Bitmap indexes are executed.
2.2410.6.0	Feature updates Supported setting default values for columns when adding a column. System optimizations Optimized slow SQL information displayed in the system table SLOW_LOG to avoid too many invalid SQL executions that may cause interference to users. Optimized the timing selection policy for executing Compact operations on backend storage. Optimized the thread volume expansion issue during concurrent execution.
2.2410.5.0	Feature updates Supported aggregate function calculation on element values fetched from JSON data types. Supported status prompt when data full loading is complete. Supported Union operations on Decimal type fields with different lengths from multiple Select statement results. System optimizations Optimized the logic of partial collection metrics during data loading, improving the accuracy of monitoring metrics.
2.2410.4.1	Bug fixes Fixed table structure inconsistency that may occur during the Add Column operation. System optimizations Improved the connection release efficiency when accessing the read-only analysis engine through Database Proxy.
2.2410.4.0	Feature updates Supported the brand-new database proxy version 1.4.4. Supported users to create a columnar storage secondary index. Bug fixes Fixed the issue where the Sort/TopN operator was unable to generate a multi-table Colocate Join. Fixed the error and alarm information when loading part of the DDL. Fixed the issue where some operators cannot be profiled. Fixed the issue where too many logs were performed during the data loading process, occupying too much storage space. Fixed the issue where the number of affected rows returned by the MySQL client and the Records entry does not correspond after the ETL writeback feature is executed.
2.2410.3.0	Bug fixes Fixed the issue where Join cannot be performed between Uint64 and Int64 field types. Fixed the issue where data loading was interrupted when encountering the Geometry data type during data loading. Fixed the issue where a parsing error occurred when creating an index on a null column. Fixed a logic issue where the table could not be reconstructed after deleting the primary key and then immediately creating a primary key during incremental data loading. System optimizations Optimized the processing policy of tables without primary key when loading data, making the loading of tables without primary key more stable.
2.2410.2.0	Feature updates Added support for the Date_Add/Find_In_Set/Weekday functions. Bug fixes Fixed the issue where excessive memory usage was caused by using Cross Join when adding a subquery in an OR expression. Fixed the issue where an error is occasionally reported when running the command if there is a constant column in SQL statements. Fixed the issue where an error is occasionally reported when running the command under an IF Expression. Fixed the issue where an execution error is reported when processing the column length in some expressions.
2.2410.1.0	Feature updates Storage features Added support for columnar storage secondary index. The supported index types include the Zonemap index, Bloom Filter index, and Bitmap index. Added support for concurrent FlushDMS operations to enhance data write efficiency, particularly improving update efficiency when managing large volumes of data. Added support for the JSON type and various JSON functions. Added support for loading range/list-partitioned tables into the LibraDB engine and loading partition DDL statements synchronously. For detailed syntax and function support related to partition operations (such as adding or dropping partitions or subpartitions), see Data Loading Restrictions. Added support for loading tables without primary keys in a columnar storage format, as well as for scenarios involving modifications to table primary keys. For details, see Data Loading Restrictions. Compute features Added support for Merge join, improving query performance in primary key join scenarios. Added support for stream aggregation, improving GROUP BY performance in ordered data scenarios. Added support for the automatic statistics collection capability, removing the need for users to manually collect statistics regularly. Added support for the random sampling feature, enhancing the efficiency of statistics collection and reducing the resource consumption associated with statistics. Added support for adaptive GROUP BY operators. Added support for streaming CTEs. For details, see CTE Syntax Use Instructions. Added support for pushing Table Dual down to the storage layer for execution. Added support for pushing the UNION ALL operator down to the storage layer for execution. Added support for the Win Magic rewrite of the optimizer, which decouples correlated subqueries by rewriting them as window functions to improve query efficiency. Added support for 14 MySQL-compatible functions. For details, see Description of Supported String Functions. Added support for string-type input values for the Hour, Minute, Second, and Microsecond functions. Added support for loading tables with decimal-type primary keys (up to a maximum length of 256) into columnar storage. Added support for NULL-aware anti-joins. When executing set operations, these joins intelligently assess whether a set is empty or contains null values, optimizing the execution efficiency of operations such as `NOT IN` and `<>ALL`. This enhancement significantly improves the performance of associated SQL queries. System optimizations Optimized the error messages returned for some execution errors. Optimized the performance of the grace hash join by employing a two-phase bucket acceleration mechanism, reducing the frequency of bucket scale-out. Added support for right anti-joins/right semi-joins in the result set splitting feature, significantly alleviating long wait times for data reception caused by extremely large join result sets. Added support for aggregation in the result set splitting feature, alleviating long wait times for data reception caused by extremely large aggregation output result sets. Optimized the EXPLAIN feature by enabling the display of the table name and alias corresponding to the TableScan operator in the detailed information of the execution plan. Optimized the EXPLAIN feature by enabling the display of replica information for the referenced table in the execution plan. Added support for scenarios where the Unique Key column contains null values. Optimized the metadata storage mechanism, significantly reducing excessive resource consumption associated with metadata. Optimized the execution method of the Apply operator, improving query performance in correlated subquery scenarios.

Version	Description
1.2404.24.0	Feature updates Supported setting default values for columns when adding a column. System optimizations Optimized slow SQL information displayed in the system table SLOW_LOG to avoid too many invalid SQL executions that may cause interference to users. Optimized the timing selection policy for executing Compact operations on backend storage. Optimized the thread volume expansion issue during concurrent execution.
1.2404.23.0	Feature updates Supported status prompt when data full loading is complete. System optimizations Optimized the logic of partial collection metrics during data loading, improving the accuracy of monitoring metrics.
1.2404.22.1	Bug fixes Fixed table structure inconsistency that may occur during the Add Column operation. System optimizations Improved the connection release efficiency when accessing the read-only analysis engine through Database Proxy.
1.2404.22.0	This version is an issue-fixed and optimized version, primarily addressing defects in versions before 1.2404.21.0 (inclusive). Bug fixes Fixed the data latency issue caused by synchronizing DDL in some scenarios. Fixed the issue where an error is reported when constructing a Datetime type for a correlated subquery column. Fixed the issue where an error is reported when the parameter is an array in the scenario of an IN expression. System optimizations Improved the performance when querying with a primary key predicate under a large data volume. Improved the truncation length of slow SQL query views when encountering relatively long SQL.
1.2404.21.0	This release is a bugfixing and optimization update that primarily addresses defects identified in version 1.2404.20.0 and earlier. Feature updates Added support for DDL operations related to partitioned tables in the current version. Bug fixes Resolved the issue that data deletion failed when the primary key contained a timestamp type. Resolved the data synchronization error associated with SET type data. Resolved several errors in the table renaming scenario.
1.2404.20.0	This release is a bugfixing and optimization update that primarily addresses defects identified in version 1.2404.19.1 and earlier. Bug fixes Resolved the issue that caused data-loading objects to be inaccessible when the read-write instance was configured as case-sensitive. Resolved the issue that some existing instances could not support database proxies. Resolved the issue that Lazy Materialization statistics were displayed incorrectly. Resolved the issue that DMC was unable to retrieve user information for the read-only analysis engine. Optimized the default values for some kernel parameter settings.
1.2404.19.1	This release is a bugfix update that primarily addresses defects identified in versions earlier than 1.2404.19.0. Bug fixes Resolved the issue of inconsistent view definers when loading views into the read-only analysis engine. Resolved the issue that user synchronization failed for users with empty passwords in the mysql.user system table. Resolved the execution error encountered when executing DDL statements during SQL operations. Fixed the issue where the "ETL Writeback Acceleration" feature encounters a writing exception when there is a constant column.
1.2404.19.0	Feature updates Added support for accessing the read-only analysis engine through the database proxy when using the HINT syntax. Added support for accessing the read-only analysis engine through DMC. System optimizations Optimized the memory management methods for the SORT and AGGREGATION operators, enhancing memory usage efficiency. Following the optimization, memory used by the SORT and AGGREGATION operators can be released immediately upon completion of their execution, without waiting for the entire query execution to end. Optimized the memory usage method for join result output, reducing memory consumption when the JOIN operator produces a relatively large result set. Optimized the allocation policy for Lazy Materialization in primary key scenarios. Bug fixes Resolved the issue that PREPARE STATEMENT defaulted to using Plan Cache, resulting in multiple similar queries inadvertently matching a previously cached plan. This could lead to query errors when the types of constant parameters were inconsistent. Resolved the issue of query errors that occurred in some scenarios when output columns of VARCHAR type were used in a UNION ALL context. Resolved the issue that metadata retrieval errors were caused when DMC connected to the columnar storage engine due to insufficient permissions. Fixed a potential memory leak issue in the scenario where joins are created frequently. Resolved the issue that temporary tables were prevented from being queried in the read-only analysis engine in scenarios involving changes to partitioned tables. Resolved the issue that the Lazy Materialization operator was not displayed in the execution plan under some scenarios. Resolved the issue that the read-only analysis engine was unable to delete fields from the join index. Resolved the out-of-bounds error in runtime filter assignment for multi-table join scenarios. Resolved the issue that the runtime filter occasionally returned empty data. Resolved the error caused by a mismatch between nullable and non-nullable types in UNION ALL scenarios.
1.2404.17.0	Feature updates Added support for the FROM_UNIXTIME and UNIX_TIMESTAMP functions. System optimizations Optimized the permissions required to query system tables when database client tools, such as DBearer, access the read-only analysis engine. Optimized error messages for some unsupported functions. Bug fixes Resolved the issue that the read-only analysis engine incorrectly loaded Timestamp field values when the explicit_defaults_for_timestamp parameter differed across instances.
1.2404.16.0	Feature updates Added support for using the read-only analysis engine to accelerate the execution of query statements in `INSERT…SELECT…`, enabling faster writes to read-write instances. For details, see ETL Writeback Acceleration. Added support for querying and synchronizing data types, including BIT, ENUM, and SET. System optimizations Improved the naming format of some system tables to make them easier to understand, such as the SLOW_LOG table. Adjusted the supported value size for regular columns to 16 MB, up from the previous limit of 5 MB. Optimized the logic of the StorageTableReader operator for retrieving column names when the query plan is empty. Bug fixes Resolved the issue that, in subquery scenarios, the Apply operator logic was still applied when the primary query result was empty, leading to a query error. Resolved the issue that the Apply operator caused a crash during multi-threaded competition.
1.2404.15.2	This release is a bugfixing update that primarily addresses defects identified in versions earlier than 1.2404.15.1. Bug fixes Resolved the issue that the instance crashed during data queries when high-concurrency `UPDATE` operations were performed on the read-write instance after it was loaded as a column-stored table.
1.2404.15.1	System optimizations Optimized the default settings for database log retention. Specifically, if an instance's disk size is 500 GB or less, the default maximum log retention is set to 5 GB; for instances with a disk size over 500 GB, it is set to 10 GB. Enabled Lazy Materialization in the read-only analysis engine by default. For details on its capabilities and tuning options, see Lazy Materialization. Added support for loading tables whose names begin with an underscore (_) into the read-only analysis engine. Improved the restart efficiency of the read-only analysis engine. Bug fixes Resolved the issue that some versions of Navicat displayed an insufficient permissions error when connecting to the read-only analysis engine. Resolved the issue that the read-only analysis engine failed to load data when DDL statements were executed consecutively. Resolved the syntax compatibility issue related to DEFAULT (0) in the `CREATE TABLE` statement.
1.2404.10.0	Feature updates Added support for the DATE_SUB function. Added support for users to manually collect table statistics. For details, see Collecting Statistics. Added support for synchronous loading of ENUM data in special scenarios. Note: Although ENUM data can be loaded into the read-only analysis engine in special scenarios, querying data of the ENUM field type is still not supported. System optimizations Adjusted the kernel to manage the number of versions retained for expired data in storage, preventing excessive disk space usage caused by retaining an excessive number of data snapshots. Optimized kernel resource allocation during data clearing, preventing degradation of user SQL execution performance resulting from excessive consumption of system resources. Adjusted the logging level for kernel information to prevent excessive log output from consuming too much disk space. Adjusted the maximum value of the storage field. The previously supported default limit of 1 MB has been increased to 5 MB. Optimized the error message for data querying when the data is not fully loaded into the read-only analysis engine. It now clearly indicates that you need to wait until all data is fully loaded before performing a query. Optimized the error message for unsupported data types and functions during execution. You can now clearly identify the unsupported items in the error message. Optimized the internal trigger frequency and selection logic for background tasks, ensuring they do not impact the execution performance of users' online SQL queries. Improved the version number information displayed when accessing the read-only analysis engine instance through the MySQL client to prevent potential misunderstandings.
1.2404.7	Feature updates Added support for real-time loading of hyperscale data into the read-only analysis engine. Added support for ultra-high performance in complex data queries. Note: This is the first launched LibraDB engine kernel version for TDSQL-C for MySQL.

tencent cloud

Version Number Rule

Version Update Description