ApsaraDB RDS:Release notes for AliSQL

Minor version

Community version

Description

Expiration date

20251231

8.0.36

• New features

  • Vector indexes can be created on nullable vector columns.

• Performance optimizations

  • The performance of lock release during transaction commits is improved.

• Bug fixes

  • The bug that the real-time DDL application feature is incorrectly applied to ALTER temporary tables is fixed.

  • The bug that causes out-of-bounds access when real-time DDL application logs are printed is fixed.

  • The performance issues caused by trx_sys lock waits in high-concurrency scenarios are fixed.

20251130

8.0.36

• New features

  • The jemalloc version is updated to 5.3.

  • A statistical item for CPU time consumption is added to audit logs.

• Performance optimizations

  • The transmission efficiency of I/O threads is optimized to improve the performance of semi-synchronous replication and the efficiency of replication on secondary instances in high-concurrency scenarios.

  • The application efficiency of SQL threads is optimized to improve the efficiency of replication on secondary instances in scenarios with a mix of medium and small transactions.

• Bug fixes

  • Bugs such as memory leaks and DDL deadlocks in vector indexes are fixed.

  • The bug that causes a crash during data archiving operations is fixed.

--

20251031

8.0.36

• New features

  • Supports vector capabilities, including new vector column types, vector operation functions, and support for creating vector indexes on vector columns.

  • The length of INT, VARCHAR, and CHAR data types can be instantly extended (instant column modification).

  • The DDL throttling feature is supported.

  • Real-time DDL application supports OPTIMIZE TABLE.

  • Real-time transmission supports setting require_row_format.

  • The cost_threshold_for_duckdb parameter is added. If the optimizer's cost estimation for a SQL command exceeds this threshold, an error is reported.

• Performance optimizations

  • The transmission efficiency of I/O threads is optimized to improve the performance of semi-synchronous replication and the efficiency of replication on secondary instances in high-concurrency scenarios.

  • Redo compression is supported for BLOB fields to improve BLOB write performance.

• Bug fixes

  • The bug that causes Seconds_Behind_Master spikes for single-statement transactions is fixed.

  • The bug that causes slow acknowledgments in semi-synchronous replication due to purging of relay logs is fixed.

  • The bug that may cause a crash when a DDL statement is executed concurrently with an UPDATE statement on a table with foreign keys is fixed.

  • The bug that causes replication interruption when the SUPER permission is granted in a native replication scenario is fixed. After the fix, only the permissions owned by the replication thread are granted.

  • The bug that causes a PREPARE statement to report an unexpected error due to a NULL value in a JSON column is fixed.

  • The performance issues caused by clearing the writeset are fixed.

--

20250731

8.0.36

• New features

  • The real-time DDL application feature is added.

  • The length of INT, VARCHAR, and CHAR data types can be extended online.

• Performance optimizations

  • The write performance for large objects is improved.

• Bug fixes

  • The bug that prevents querying implicit primary keys from information_schema.columns after an upgrade from an earlier version is fixed.

  • The bug that causes unexpected log records when an index is created and innodb_log_optimize_ddl is set to OFF is fixed.

  • The community bug #118476, which prevents creating partitions with certain identifiers, is fixed. For more information, see Bug #118476.

  • The community bug #118872, which causes inaccurate statistics for Innodb_data_written, is fixed. For more information, see Bug #118872.

  • The community bug #105652, which causes an exception during crash recovery for large tables that exceed 32 TB, is fixed. For more information, see Bug #105652.

  • The community bug #117275, which may cause incorrect query results when index_merge uses a descending primary key index, is fixed. For more information, see Bug #117275.

--

20250531

8.0.36

• New features

  • CURL is updated to version 8.14.1.

  • Real-time transmission of large transactions to secondary instances during execution is supported. This avoids performance fluctuations caused by committing large transactions on semi-synchronous replication instances.

  • New hint syntax (WAIT N and NO_WAIT) is supported. DML and DDL statements can set the MDL wait time using hints.

  • Non-blocking DDL statements are supported to prevent DML statements from being blocked due to MDL lock timeouts during DDL execution.

  • A lightweight performance_schema can be used to view the execution status of DDL statements and the status of MDL locks.

  • Fast parsing of row records is supported for tables on which INSTANT DDL has not been performed.

  • Binlog in redo and Binlog Parallel Flush can be enabled in semi-synchronous replication mode.

  • The end time of SQL statements can be recorded in the start_time field of the slow query log.

  • A performance optimization feature for hot row updates is added.

  • The issue where DDL execution fails due to unique key conflicts is optimized.

  • Information about the number of active transactions and the average transaction execution time is added to the Innodb status.

• Performance optimizations

  • The performance of primary key scanning, sorting, and index building during DDL execution is improved.

  • The parsing performance of the utf8mb3 character set is improved.

• Bug fixes

  • The bug that causes slow execution of reset slave all is fixed.

  • The bug that the binlog_transaction_compression_level_zstd parameter does not take effect is fixed.

  • The bug that causes a crash when updating a table with an instant drop column is fixed.

  • The bug in the writeset feature for foreign keys, which could cause replication interruption when an INSERT statement does not specify a value for an auto-increment column, is fixed.

  • The bug that causes long upgrade times when the General Query Log is enabled is fixed.

  • The bug that may cause incorrect Hash Anti Join results when join_buffer_size is small is fixed.

  • The bug that optimize table does not update the table status is fixed.

  • The bug that causes a statement id overflow is fixed.

  • The bug that causes the SEQUENCE engine to crash is fixed.

--

20241231

8.0.36

• New features

  • CURL is updated to version 8.11.1.

  • The column encryption feature supports KMS-based key management.

  • The column encryption feature supports the AES 256 algorithm.

• Performance optimizations

  • Multi-threaded parallel replay of redo logs is supported to accelerate instance startup during instance crash recovery.

• Bug fixes

  • The bug that the table status becomes incorrect due to rename operations is fixed.

  • The bug that the server IDs in the error log are incorrect is fixed.

  • The bug that the value of the start_time field is incorrect in the stored procedure calling information in the slow query log is fixed.

  • The bug that the undo tablespace information in information_schema.files is inaccurately displayed after MySQL restart is fixed.

  • The bug that the automatic truncation of the undo tablespace fails during MySQL restart is fixed.

  • The bug that error logs related to stored procedures, events, and triggers are missing during major engine version upgrades is fixed.

  • The bug that memory leaks during major engine version upgrades is fixed.

  • The bug that garbled characters are not checked during the creation or modification of events and triggers is fixed.

--

20241130

8.0.36

• New features

  • OpenSSL is updated to version 3.0.15.

  • curl is updated to version 8.9.1.

• Performance optimizations

  • The period of time that is required for the automatic remediation of the General Query Log is reduced to improve the RTO of the instance.

  • The scanning of double-write operations during instance recovery is optimized to reduce the RTO of the instance. For more information, see Community Bug #116209.

• Bug fixes

  • The bug that the instance crashes due to tables archived in OSS is fixed.

  • The bug that the ciphertext in the EncDB column collides is fixed.

  • The bug that binary logs are recorded for failed optimization is fixed.

  • The bug that the instance becomes unavailable because the buffer pool is used up by internal temporary tables is fixed. For more information, see Community Bug #96236.

  • The bug that the instance may crash when the buffer pool size is reduced is fixed.

  • The bug that the instance crashes or invalid query results are returned when the common table expression (CTE) column is referenced and the CTE table is materialized is fixed.

  • The bug that the instance crashes when the view definition includes a subquery and the user who defines the view does not exist or have required permissions is fixed.

  • The bug that the instance may crash when multi-value indexes are used is fixed.

  • The bug that the instance crashes when ranking is performed in full-text searches is fixed.

--

20241031

8.0.36

• Bug fixes

  • The bug that data may be lost during the execution of an INPLACE DDL statement is fixed. For more information, see Community Bug#115608.

--

20240930

8.0.36

• New features

  • The innodb_instant_ddl_enabled parameter is added to control whether to enable instant DDL.

• Bug fixes

  • The bug that causes the instance to crash when you perform an UPDATE operation on a column added using an instant DDL statement is fixed.

  • The bug that causes the instance to crash when you perform a DELETE or UPDATE operation on a column deleted using an instant DDL statement is fixed.

--

20240731

8.0.36

• New features

  • The empty events that are generated by the binlog cache free flush feature can be skipped by the binary log dump thread.

• Performance optimizations

  • The global transaction identifier (GTID) allocation is optimized to improve performance in high concurrency scenarios.

  • The performance issue of the MD5 algorithm in high-concurrency scenarios is fixed.

  • The issue that the RDS instance fails to be connected because of the automatic remediation of the General Query Log is fixed.

  • The process to obtain and verify the tablespace metadata during instance startup is optimized. This helps accelerate the startup of RDS instances with many tables.

  • The duration of a lock that occurs when you execute the DROP TABLE or TRUNCATE TABLE statement is shortened. The innodb_rds_drop_ahi_ahead parameter is supported. If you enable the parameter, the adaptive hash indexes created for InnoDB tables are deleted before you drop or truncate the InnoDB tables. This shortens the duration of the metadata lock on the InnoDB tables during the execution of the DROP TABLE or TRUNCATE TABLE statement and prevents other SQL statements from being blocked.

• Bug fixes

  • The bug due to which the system that has the parser component does not respond during the creation of full-text indexes is fixed.

  • The bug due to which the system does not respond when NULL parameters are passed to a plug-in after EncDB is enabled is fixed.

  • The bug due to which the system does not respond when tables are archived to an Object Storage Service (OSS) bucket is fixed.

  • The bug due to which page caches are consumed during DDL execution is fixed.

  • The bug due to which the number of updated rows in audit logs is inaccurate when INSERT and UPDATE statements fail to be executed is fixed.

  • The bug due to which the system does not respond when the ALTER TABLE statement is executed to add an auto-increment column as a primary key in descending order is fixed.

  • The bug due to which binary logs are recorded even if the OPTIMIZE TABLE statement fails to be executed is fixed.

--

20240630

8.0.36

• Bug fixes

  • The bug of the unexpected internal permission management of system tables is fixed to allow the TRUNCATE operation on tables in Performance Schema using a user account.

  • Fixed a potential memory leak that occurred during redo log processing for table records after an INSTANT DDL operation.

--

20240615

8.0.36

• Bug fixes

  • The bug that redo logs fail to be written because the sequence of non-INSTANT columns are changed for the instant DDL feature is fixed.

--

20240531

8.0.36

• New features

  • The updates to the features in MySQL 8.0.35 and 8.0.36 are incorporated.

  • The NET::buff memory for persistent connections is automatically released to reduce the use of memory resources and reduce the risks of out-of-memory (OOM) errors.

• Performance optimizations

  • Asynchronous I/O reads are supported by Buffer Pool Extension.

  • The process to scan and verify data files during the instance startup is optimized to improve the startup speed of an RDS instance that has many tables.

• Bug fixes

  • Fixed the issue of incorrect timestamps in GTID Log Event generated by Binlog Cache Free Flush.

  • The following issue is fixed: If a column named FTS_DOC_ID is explicitly created in an InnoDB table, the major engine version upgrade from 5.7 to 8.0 fails.

  • The bug due to which data types returned by EncDB are disordered is fixed.

  • The following issue is fixed: After the recycle_bin parameter is disabled, read-only instances still trigger a reclaim.

  • The bug that innodb_activity_count may be negative is fixed. For more information, see Community Bug #114180.

  • The following issue is fixed: The parallel read partition table causes the parallel thread count to leak and causes parallel read failures. For more information, see Community Bug #114154.

--

20240522

8.0.34

• Bug fixes

  • The bug that redo logs are disordered because non-Instant columns are changed for the Instance DDL feature is fixed.

--

20240229

8.0.34

• New features

  • OpenSSL is updated to version 3.0.12.

  • The fault diagnostics logging feature is supported to record DDL operations.

  • Adds the Confidential database (public preview) feature to implement end-to-end data encryption and protect data security.

• Bug fixes

  • The bug due to which semi-synchronous replication fails when an error occurs in the comparison of binary log file names is fixed.

  • The bug that compatibility issues of the backup set occur due to IPK is fixed.

--

20240131

8.0.34

• Bug fixes

  • The bug that the memory of the performance_schema.memory_summary_by_thread_by_event_name table is overcommitted after the thread pool feature is enabled is fixed.

  • The bug due to which the duplicate entry error frequently occurs in the auto-increment column after an RDS instance using X-Engine is restarted is fixed.

  • The bug due to which the "ERROR 1878 (HY000): Temporary file write failure." error message is displayed when the innodb_disable_sort_file_cache=ON setting is used and a full-text index is created is fixed.

  • The bug due to which the full-text index parser recreated in a statement using the INPLACE DDL statement is invalid is fixed. For more information, see Community Bug #110976.

  • The bug that the auto cleanup feature for read-only RDS instances cannot be used as expected after the recycle_bin parameter is modified is fixed.

  • Fixed the issue where executing CREATE SEQUENCE in a PREPARE statement caused the instance to crash.

--

20230930

8.0.34

• New features

  • The updates to the features in MySQL 8.0.32, 8.0.33, and 8.0.34 are incorporated.

  • OpenSSL is upgraded from version 1.1.1u to 3.0.10.

  • The built-in function SM3() is added.

  • Buffer Pool Extension supports Online Resize.

  • The default value of the ssl_cipher parameter is changed to "ALL:@SECLEVEL=0".

• Performance optimizations

  • The binlog cache free flush feature is supported.

    During the transaction commit phase, this feature converts the temporary binary log cache files of large transactions into binary log files. This reduces the time that the global binlog lock is held when large transactions are committed, reduces the I/O pressure on the RDS instance, and prevents the workload switchover of the RDS instance. The workload switchover is triggered because the binary logs cannot be written to the RDS instance or I/O hangs occur on the instance for a long period of time.

  • The replication latency of 1s is optimized.

  • The memory used by the buffer pool extension feature is optimized.

  • The I/O efficiency of parallel reads is optimized when innodb_use_native_aio is set to off. For more information, see Community Bug #112137.

• Bug fixes

  • The bug due to which the pfs metric does not collect statistics on the buffer pool and buffer pool extension is fixed.

  • The bug due to which the major engine version cannot be upgraded because the case sensitivity of the names of fields in indexes at the server layer is inconsistent with that at the InnoDB layer is fixed.

--

20230914

8.0.31

• Bug fixes

  • When you add or remove fields, the operation defaults to INPLACE instead of INSTANT to prevent stability issues associated with INSTANT.

--

20230630

8.0.31

• New features

  • The updates to the features in 8.0.31 are incorporated.

  • The in-progress online buffer pool resizing can be canceled.

  • The template-level concurrency control is supported by the statement concurrency control feature. This feature lets you limit the execution of SQL statements in the same template. For more information, see SQL throttling.

  • zlib is updated to version 1.2.13.

  • OpenSSL is updated to version 1.1.1u.

  • The Innodb_redo_file_used_ratio status item is added to show the ratio of used redo files.

  • X-Engine supports deployment in ARM environments.

• Performance optimizations

  • mysqld_safe.sh is optimized to accelerate instance startup.

• Bug fixes

  • The bug due to which XA transactions cannot be correctly recovered after an RDS instance is restarted is fixed. For more information, see Community Bug #110533.

  • The bug that an RDS instance is unexpectedly restarted because sql_type for the thread pool is not correctly configured is fixed.

  • The bug due to which debugging information is lost in the installation package is fixed. For more information, see Community Bug #109560.

  • Fix the MySQL 5.7 to MySQL 8.0 upgrade failure caused by temporary tables created in MySQL 5.6.

  • The bug due to which X-Engine transactions are occasionally lost in deadlock scenarios is fixed.

  • The bug due to which transactions cannot be committed when Binlog in Redo and Binlog Parallel Flush work under heavy workloads for a long period of time is fixed.

--

20230620

8.0.28

• Bug fixes

  • The group replication delayed startup feature prevents cluster operations from failing to execute when instances restart and reconnect to the cluster quickly.

  • The bug due to which transactions cannot be committed when Binlog in Redo and Binlog Parallel Flush work under heavy workloads for a long period of time is fixed.

--

20230610

8.0.28

• Bug fixes

  • The issue due to which deadlocks occur when you execute the COM_STATISTICS, COM_CHANGE_USER, or SHOW PROCESSLIST statement is fixed. The deadlocks cause the issue that your RDS instance cannot be connected. For more information, see Community Bug #110494.

  • The issue due to which backup tasks cannot be executed because the background reclamation tasks of X-Engine unexpectedly exit is fixed.

--

20230324

8.0.28

• Bug fixes

  • The bug that the Binlog In Redo feature is unexpectedly enabled in semi-synchronous replication mode is fixed.

--

20221231

8.0.28

• New features

  • The parallel write feature can be enabled to write binary logs when the value of the sync_binlog parameter is not 1. After the parallel write feature is enabled, binary logs can be concurrently written to your RDS instance. This improves the write performance of your RDS instance.

  • In single leader mode, MGR can be used to forcefully delete authentication information. By default, the feature for forcefully deleting authentication information is enabled. You can disable the feature. This feature helps you resolve the issues such as periodic performance jitters, high memory usage, and complete performance loss due to new nodes in a cluster in high-concurrency and high-latency business scenarios.

  • When the SQL statement for MGR to proactively select the primary node is initiated, if the execution duration of a transaction exceeds the timeout period that is configured for the SQL statement, the transaction is forcefully rolled back. This prevents primary node selection from being blocked and improves the success rate of primary node selection.

  • OpenSSL is updated to version 1.1.1s.

• Performance optimizations

  • The Purge Large File Asynchronously feature is enhanced for ApsaraDB RDS for MySQL. This accelerates the execution of the DROP TABLE and TRUNCATE TABLE statements for tables that have many dirty pages.

  • The purge statistical items are added to Performance Agent. The following list describes the statistical items.

    • INNODB_UNDO_SIZE_USED: the size of used undo logs

    • INNODB_PURGE_LIMIT_TRX_NO: the purge offset, which indicates the largest transaction number among purged transactions

    • INNODB_PURGE_ITER_TRX_NO: the maximum number of transactions that can be purged

• Bug fixes

  • The bug that may cause the instance to crash when EXPLAIN UPDATE involves a derived table in a subquery is fixed. For more information, see Community Bug#31884434.

  • The bug due to which the execution result of the SELECT COUNT(*) statement is inaccurate when multi-value indexes are used is fixed. For more information, see Community Bug #104898.

  • The bug due to which only partial result sets are returned for a query after you add multi-value indexes to tables with JSON columns is fixed. For more information, see Community Bug #106621.

  • The bug due to which a memory leak is caused by the Clone_persist_gtid thread is fixed. For more information, see Community Bug #107991.

  • The bug due to which the number of updated rows in the audit logs is inaccurate when the UPDATE and INSERT statements are executed and the CLIENT_FOUND_ROWS flag is enabled is fixed. By default, the CLIENT_FOUND_ROWS flag is disabled.

  • The bug due to which the major engine version cannot be upgraded from MySQL 5.7 to MySQL 8.0 because the names of table columns at the server layer and the InnoDB layer do not match is fixed. The names of table columns at the server layer and InnoDB layer are case-sensitive.

  • The bug due to which the major engine version cannot be upgraded from MySQL 5.7 to MySQL 8.0 because invalid strings exist in tables, fields, and index comments is fixed.

--

20221121

8.0.28

• Bug fixes

  • The bug due to which the reference to a temporary table cannot be properly applied to a common table expression (CTE) when you perform a query using the CTE in some scenarios is fixed.

--

20220830

8.0.28

• New features

  • The updates to the features in MySQL 8.0.26, 8.0.27, and 8.0.28 are incorporated.

  • The json_document_max_depth parameter is supported to configure the maximum depth that is allowed when JSON data is parsed.

• Performance optimizations

  • The parallel initialization performance of the buffer pool is optimized.

• Bug fixes

  • The bug that redo logs are recorded in a temporary tablespace is fixed.

--

20220730

8.0.25

• Performance optimizations

  • The Buffer Pool Free Resize feature is optimized.

--

20220530

8.0.25

• New features

  • If your database system consists of only one RDS instance that uses cloud disks, redo logs can be used to perform incremental backups and restore data to any point in time.

  • The Persist Binlog Into Redo V2 feature is supported. High concurrency capabilities are optimized to deliver better performance.

  • The SSL encryption feature is optimized, and a few bugs of the feature are fixed.

  • The permissions to manage the sys or performance_schema database are revoked from standard accounts.

  • The FLUSH TABLES WITH READ LOCK and UNLOCK TABLES logs are added to error logs, which facilitates online troubleshooting.

• Performance optimizations

  • The performance of buffer pool initialization is improved.

  • During the recovery from a system downtime, the rollback method of in-doubt transactions is changed from synchronous rollback to asynchronous rollback to improve the recovery time objective (RTO).

  • The memory usage and update speed of the adaptive hash index (AHI) is optimized.

  • Bidirectional replication is optimized. When a secondary RDS instance sends binary logs to its primary RDS instance, the secondary RDS instance skips transactions that belong to the primary RDS instance. This reduces the network traffic of circular replication.

  • The time to recover from an instance crash is reduced.

• Bug fixes

  • The bug that the mysql.events table fails to be migrated during an upgrade from MySQL 5.7 to MySQL 8.0 is fixed.

  • The bug that may cause a system crash when a full-text index table is deleted during an upgrade from MySQL 5.6 to MySQL 8.0 is fixed.

  • The bug that a system crash may occur due to high metadata lock (MDL) competition is fixed.

  • The bug that causes xengine_record parsing errors is fixed.

  • The bug that causes compression failures when X-Engine creates indexes is fixed.

  • The data in the start_time column of the slow_log table is amended.

  • The compilation speed is improved. Compilation warnings are fixed.

--

20220523

8.0.25

• BUG fixes

  • The bug that the execution speed of the SELECT COUNT(*) statement without the WHERE clause is slow is fixed.

  • The bug that the system fails because the stored procedures with the WITH ROLLUP syntax are called multiple times is fixed.

--

20220331

8.0.25

• New features

  • Removes the restriction that innodb_buffer_pool_instances cannot exceed 1 when innodb_buffer_pool_size is less than 1 GB.

  • Reduces the impact on instance performance when the Native Flashback feature is enabled.

• Bug fixes

  • The bug due to which the Unknown thread id message is displayed when a user terminates a session is fixed.

  • The bug that the settings of the rds_expose_priv_list parameter do not take effect after an RDS instance restarts is fixed.

  • Fixed the issue where the Binlog in Redo feature blocks checkpoint offset advancement.

  • The fixes for Bug#33341080 and Bug#32962511 in MySQL 8.0.25 of MySQL Community Edition are incorporated.

--

20210930

8.0.25

• New features

  • The updates to the features in MySQL 8.0.25 are incorporated.

  • The recovery_apply_binlog mechanism is provided. This mechanism enables XA transactions to support the Crash Safe feature.

  • The audit log feature is upgraded to version 3. In version 3, a column is added to store transaction IDs, and a field named ext is added to store the identifiers that are used to mark where SQL statements are truncated.

  • A new sequence can be defined as a timestamp sequence. The format of a timestamp sequence is different from the format of a digital sequence. Digital sequences are an outdated type of sequence.

    Syntax: CREATE SEQUENCE seq CACHE [Cache size] TIMESTAMP;

  • The truncate permission on the mysql.slow_log table and the mysql.general_log table is granted to privileged accounts.

  • The Native Flashback Query feature is supported. This feature allows users to query the data that is rolled back and restore the data using SQL statements.

  • A proprietary X-Tree structure is provided. This structure is used to create indexes in X-Engine MemTable. This structure also supports writes, point queries, and range queries at high performance.

  • The size of the buffer pool can be adjusted. The adjustment process is optimized to prevent impacts on instance performance.

  • The Multi Blocks Read feature is supported. You can add the /*+ MULTI_BLOCKS(n) */ hint to SQL statements to prefetch multiple data pages.

  • The logic that is used to scan the linked list of the buffer pool and obtain idle pages is optimized. The least recently used (LRU) algorithm is used to manage the linked list.

  • The UPDATE and DELETE statements that contain a primary key or a unique key can automatically enter a CCL queue.

  • Add the TCP write wait duration to the performance_schema.events_statements_summary_by_digest_supplement table.

  • DDL operations that are performed using the In Place method can be recorded in the redo log.

  • By default, MyISAM is added to the value of the disabled_storage_engines parameter to prevent you from creating MyISAM tables.

• Bug fixes

  • The bug due to which the execution result of the SHOW GLOBAL STATUS statement is abnormal is fixed.

  • The bug due to which an RDS instance cannot start when the profiling feature of Jemalloc is enabled for the RDS instance is fixed.

  • The bug due to which an RDS instance with the sequence feature is enabled unexpectedly exits when the INSERT INTO SELECT statement is executed is fixed.

  • The dbms_recycle.restore_table permission is revoked.

  • The metric that specifies the number of index scans is supported for the I_S.INDEX_STATISTICS table.

  • More metrics that are used to measure the performance of an RDS instance are supported for the I_S.PERF_STATISTICS table.

  • Optimize the call performance of user-defined functions (such as nextval and currval) when they conflict with the SEQUENCE system function.

  • X-Engine can immediately respond to KILL SESSION requests.

  • The bug that causes inconsistency errors if specific DDL operations fail in extreme circumstances is fixed.

  • A stack overflow protection mechanism is added.

  • The bug that prevents the mysqld program from starting when persist_binlog_to_redo and X-Engine are used at the same time is fixed.

  • The bug that causes memory leaks when the pthread_getattr_np function is called is fixed.

  • The page cache generated by Performance Agent files is automatically cleared.

  • Fixed an internal error in the Statement Outline feature.

  • The bug due to which the system resets the used session tracker when errors occur during the execution of statements is fixed.

  • The ccl_wait_timeout variable is renamed.

    Note

    ccl_wait_timeout specifies the maximum wait time in seconds for a throttled statement.

  • The bug due to which an instance restart fails because socket files are not deleted when the database process stops is fixed.

  • The priorities of the user-defined functions (nextval or currval) are adjusted.

  • The bug due to which the net_length_size function returns error 251 is fixed.

--

20201031

8.0.18

• New features

  • You can restore tables from the Recycle Bin.

  • The slow query log files of an RDS instance can be automatically restored when the instance is initialized.

• Performance optimizations

  • You cannot enable Binlog in Redo if you use the X-Engine engine.

• Bug fixes

  • The bug that causes errors during the execution of the ASSERT statement when the value of the key for the unique index is abnormally large is fixed.

  • The bug that prevents the COM_DAEMON process from stopping is fixed.

  • The bug that causes a cache overflow when Full-Text Search (FTS)-based queries are run is fixed.

  • The bug that causes errors in rollbacks when instant DDL operations become abnormal is fixed.

20240601

20200831

8.0.18

• New features

  • Adds an option to specify whether to allow the COUNT(*) function to run scans in parallel. If the innodb_parallel_read_threads parameter is set to 0, this function cannot run scans in parallel. For more information, see Set instance parameters.

  • Start GTID (start gtid) and end GTID (stop gtid) are supported by mysqlbinlog.

  • LSNs are supported in the redo log:

    • innodb_lsn: the LSN of each record in the redo log.

    • innodb_log_checkpoint_lsn: the LSN of the last checkpoint.

    • The innodb_log_write_lsn represents the LSN for log writes.

    • innodb_log_ready_for_write_lsn: The Log Sequence Number (LSN) up to which the log buffer is ready for writing.

    • innodb_log_flush_lsn: The LSN of the redo log flushed to disk.

    • innodb_log_dirty_pages_added_up_to_lsn: The LSN that indicates the limit of dirty page additions.

    • innodb_log_oldest_lsn: the LSN (Log Sequence Number) for page flushing.

• Performance optimizations

  • Optimize the wait and concurrency mechanisms for Concurrency Control (CCL).

  • You can adjust the execution priority of Concurrency Control in stored procedures.

• Bug fixes

  • The bug due to which the recursively called parser does not check the heap memory size is fixed.

  • The bug due to which table definitions cannot be modified when transparent data encryption (TDE) is enabled is fixed.

  • The bug due to which the event scheduler causes memory leaks is fixed.

20240601

20200630

8.0.18

• New features

  • Faster DDL: optimizes the buffer pool management mechanism during DDL operations, reduces the performance impact of DDL operations, and increases the concurrency of online DDL operations.

  • The maximum number of connections that are allowed is increased to 500,000.

• Performance optimizations

  • The thread pool feature is optimized.

  • Set the upper memory limit for Performance Schema according to the instance type.

  • Audit log files are no longer analyzed.

  • TDE is optimized to cache the keys that are provided by Key Management Service (KMS).

  • Modified the status of threads running in SQL throttling.

• Bug fixes

  • Fixed an issue where semicolons (;) were incorrectly included as part of the input query during outline summary calculation.

  • The bug due to which an RDS instance unexpectedly exits when tables are modified is fixed.

  • The bug due to which earlier versions do not support the keywords that are supported in later versions is fixed. The keywords are member and array.

  • The bug due to which the system incorrectly counts the number of waits when commands are read from a database client is fixed.

  • The bug that causes failures in updating the minor engine version of an RDS instance is fixed.

20240601

20200430

8.0.18

• New features

  • Binlog in Redo: Optimizes the transaction persistence mechanism by writing binary log records to the redo log and improves database performance.

  • The row cache code of the X-Engine engine is refactored.

  • The XA_RECOVER_ADMIN permission is available.

• Performance optimizations

  • When you operate on InnoDB temporary tables, only the dirty page list is scanned, not the entire buffer pool list.

  • For compatibility with MySQL 5.6, the global parameter opt_readonly_trans_implicit_commit is renamed to rds_disable_explicit_trans.

  • During an instance upgrade, upgrade-related logs are not recorded in the audit log.

  • The memory consumed by DDL operations on X-Engine tables is reduced.

• Bug fixes

  • The bug due to which the actual size of X-Engine tables on the disk is inconsistent with the statistics in the IS table is fixed.

  • The bug due to which the system initializes X-Engine logs when the error log file is re-opened is fixed.

20240601

20200331

8.0.18

• New features

  • The updates to the features in official MySQL 8.0.17 and 8.0.18 are incorporated.

  • Recycle Bin: adds support for the TRUNCATE TABLE command. When you execute this command, the original table is moved to a dedicated recycle bin folder, and a new table with the same structure is created.

• Performance optimizations

  • The output of TCP errors is disabled by default.

  • The thread pool feature is optimized to provide higher performance when the default configurations are used.

• Bug fixes

  • The bug that causes invalid databases and tables is fixed. This bug is triggered when the names of partitioned tables are separated by the combination of a pound key and a letter p (#p).

  • The bug due to which CCL-managed statements are case-sensitive is fixed. Now, they are not case-sensitive.

20240601

20200229

8.0.16

• New features

  • Performance Agent is a MySQL plugin that collects and analyzes performance data from a MySQL instance to simplify statistical analysis.

  • In semi-synchronous mode, the network round trip time is recorded in the performance data.

  • Online DDL operations are supported by X-Engine.

• Performance optimizations

  • You can perform statement concurrency control (CCL) operations on read-only instances.

  • Outline is supported on standby instances.

  • The database proxy feature is enhanced to optimize short-lived connections.

  • The amount of time that is required to execute the pause instruction is reduced in various CPU architectures.

  • A memory summary table is added to provide the statuses of thread pools.

  • Thread pool performance optimization.

• Bug fixes

  • For Linux kernels older than 4.9, disable ppoll and use poll instead.

  • The bug that causes errors when the wrap_sm4_encrypt function is invoked is fixed.

  • The bug due to which the system locks global variables during the rotation of audit logs is fixed.

  • The bug that causes errors during restoration inconsistency checks is fixed.

  • The bug that causes inaccurate time values in the io_statistics table is fixed.

  • The bug due to which an RDS instance unexpectedly exits when invalid compression algorithms are used is fixed.

  • The bug that causes user columns in MySQL 8.0 and MySQL 5.6 to be incompatible is fixed.

  • The bug that causes buffer count leaks is fixed.

20240601

20200110

8.0.16

• New features

  • Inventory Hint: Three new hints have been added, supporting SELECT, UPDATE, INSERT, and DELETE statements to quickly commit or roll back transactions and improve business throughput capacity.

• Performance optimizations

  • When you start an instance, the system first initializes the Concurrency Control queue structure and then the Concurrency Control rules.

  • Unlinks small files during asynchronous file purging.

  • Optimized Thread Pool performance.

  • The restoration inconsistency check feature is disabled by default.

  • Permissions required to change setting variables:

    • The permissions required to set the following variables have been changed to regular user permissions:

      • auto_increment_increment

      • auto_increment_offset

      • bulk_insert_buffer_size

      • binlog_rows_query_log_events

    • The permissions required to set the following variables have been changed to superuser or system variable administrator permissions:

      • binlog_format

      • binlog_row_image

      • binlog_direct

      • sql_log_off

      • sql_log_bin

20240601

20191225

8.0.16

• New features

  • Recycle Bin: You can temporarily transfer deleted tables to the recycle bin and set a retention period so you can recover your data.

• Performance optimizations

  • The mechanism that is used to process short-lived connections is optimized.

  • A dedicated thread is used for the maintain user service. This prevents HA failures.

  • The locking mechanism is optimized. If an error occurs when the redo mechanism is used to flush binary log records, the system can explicitly release the locks that are triggered by file synchronization.

  • Unnecessary TCP error logs can be deleted.

  • The thread pool feature is enabled by default.

• Bug fixes

  • The bug that causes errors during the updates of slow query logs is fixed.

  • The bug that causes an inappropriate lock scope is fixed.

  • The bug that causes errors in core dumps when the Select function is called for TDE is fixed.

20240601

20191115

8.0.16

• New features

  • Statement Queue: A queuing mechanism for statements that uses bucket queue to group statements that may have the same conflicts into the same bucket, reducing conflict overhead.

20240601

20191101

8.0.16

• New features

  • The SM4 encryption algorithm is supported for TDE.

  • Protect backup instance information: Only users with the SUPER or REPLICATION_SLAVE_ADMIN privilege can insert, delete, or modify data in the slave_master_info, slave_relay_log_info, and slave_worker_info tables.

  • A mechanism is provided to increase the priorities of auto-increment keys. If a table does not have a primary key or the table does not have a unique key without a null value, the auto-increment key without a null value is assigned the highest priority.

  • A mechanism is provided to prevent the automatic conversion of tables from the Memory engine to the MyISAM engine. These tables include system tables and the tables that are called by threads in the initializing state.

  • A mechanism is provided to flush binary log records to the disk before redo log records.

  • A locked instance also affects temporary tables.

  • X-Engine that is based on a log-structured merge (LSM) tree is provided to store transactions.

• Performance optimizations

  • Thread Pool: Optimization to reduce mutual exclusions (mutexes).

  • Performance Insight: The performance insight feature supports thread pools.

  • The following parameters are adjusted:

    • primary_fast_lookup: a session parameter. Default value: true.

    • thread_pool_enabled: a global parameter. Default value: true.

20240601

20191015

8.0.16

• New features

  • TDE: The transparent data encryption (TDE) feature performs real-time I/O encryption and decryption on data files. Data is encrypted before it is written to disk and decrypted when it is read from the disk into memory.

  • Returning: The Returning feature allows DML statements to return a result set and provides the DBMS_TRANS package to help you use this feature easily.

  • Forced conversion from the MyISAM or MEMORY storage engine to the InnoDB engine is supported. If the global variable force_mysiam_to_innodb or force_memory_to_innodb is set to ON, tables are automatically converted from the MyISAM or MEMORY storage engine to the InnoDB engine when the tables are created or modified.

  • A mechanism is provided to prevent accounts that are not privileged accounts from performing primary/secondary switchovers. Only privileged accounts are granted the permissions to perform primary/secondary switchovers.

  • A performance agent plugin is provided. This plugin obtains performance data and saves the data as text files to on-premises computers. A round-robin algorithm is used to retain only the most recent files that contain performance data at the single-digit second level.

  • A timeout period can be configured for mutexes in InnoDB. You can change the timeout period using the global variable innodb_fatal_semaphore_wait_threshold. The default value of the global variable is 600.

  • Index hint errors can be ignored. You can configure an RDS instance to ignore index hint errors using the global variable ignore_index_hint_error. The default value of the global variable is false.

  • You can disable the SSL encryption feature.

  • The output of TCP errors is supported. TCP read, read-wait, and write-wait errors are returned together with error codes in end_connection events. Log records that contain the information about the errors are also generated.

• Bug fixes

  • In Linux systems that support local AIO, AIO requests are merged before linear read-ahead is triggered.

  • Optimize table and index statistics.

  • If a primary key is specified, you can directly access the primary index.

20240601

20190915

8.0.16

• Bug fixes

  • The bug that causes memory leaks when the Cmd_set_current_connection process runs is fixed.

20240601

20190816

8.0.16

• New features

  • Thread Pool: Separates threads from sessions. This allows the system to handle many sessions using only a small number of threads to complete tasks for active sessions.

  • SQL throttling: By controlling concurrency, SQL throttling helps handle traffic bursts, statements that consume excessive resources, and changes in SQL access patterns, ensuring the continuous and stable operation of MySQL instances.

  • Statement Outline: Use Optimizer Hints and Index Hints to stabilize MySQL execution plans.

  • Sequence Engine: simplifies the generation of sequence values.

  • Purge Large File Asynchronously: When you delete a single tablespace, the system renames the tablespace files as temporary files and waits for the asynchronous purge process to clean up the temporary files.

  • Performance Insight: A powerful tool for instance load monitoring, association analysis, and performance tuning that helps you quickly assess database load, identify the root cause of performance issues, and improve database stability.

  • The instance lock status is optimized. If an RDS instance is locked, the DROP or TRUNCATE operation can still be performed on tables.

• Bug fixes

  • The bug due to which the system incorrectly calculates file sizes is fixed.

  • Fixed an occasional issue where memory was reused after becoming idle.

  • The bug due to which a host unexpectedly exits when the available cache size on the host is 0 is fixed.

  • The bug that causes conflicts between implicit primary keys and CTS statements is fixed.

  • The bug due to which the system incorrectly generates logs for slow queries is fixed.

20240601

20190601

8.0.16

• Performance optimizations

  • The MDL scope for log tables is shortened to reduce the possibility of MDL blocking.

  • The code for termination options is refactored.

• Bug fixes

  • The bug due to which precompiled statements are not recorded by the audit log feature is fixed.

  • You can suppress error logs for invalid table names.

20240601

Minor version

Description

20230415

• Bug fixes

  • Fixed a stability bug that occurred in large transaction scenarios.

  • The bug that the value of the Seconds_Behind_Master parameter is not 0 when no traffic exists is fixed.

20230406

• Performance optimizations

  • The faster DDL performance is optimized. When a session exits, if temporary tables are generated by user operations or the optimizer, the default process to evict pages is optimized from traversing the entire buffer pool to traversing only pages related to the temporary tables. This reduces the waiting time for the buffer pool to evict temporary tables, reduces conflicts, and improves the overall performance.

20210305

• Bug fixes

  • The bug that causes memory leaks for performance schema is fixed.

20200918

• New features

  • SSL encryption can be enabled for the dedicated proxy.

20200805

• Bug fixes

  • A metadata-related bug is fixed.

20200608

• New features

  • The updates to the features in official MySQL 8.0.17 and 8.0.18 are incorporated.

  • Recycle Bin: Adds support for TRUNCATE TABLE command. When executed, the original table is moved to a dedicated recycle bin folder, and a new table with the same structure is created.

• Performance optimizations

  • The output of TCP errors is disabled by default.

  • The thread pool feature is optimized to provide higher performance when the default configurations are used.

• Bug fixes

  • The bug that causes invalid databases and tables is fixed. This bug is triggered when the names of partitioned tables are separated by the combination of a pound key and a letter p (#p).

  • The bug due to which CCL-managed statements are case-sensitive is fixed. Now, they are not case-sensitive.

20200317

• New features

  • Performance Agent: A more convenient solution for performance data statistics. This feature is provided as a MySQL extension and lets you collect and analyze various performance metrics for your MySQL instance.

  • Faster DDL: Optimizes the Buffer Pool management mechanism during DDL operations, which reduces the performance impact and increases the concurrency of online DDL operations.

  • In semi-synchronous mode, the system includes network round trip time in the performance data.

• Performance optimizations

  • The CCL feature is optimized to control the concurrency of statements on read-only RDS instances.

  • Outline is supported on standby instances.

  • The database proxy feature is enhanced to optimize short-lived connections.

  • The amount of time that is required to execute the pause instruction is reduced in various CPU architectures.

  • A memory summary table is added to provide the statuses of thread pools.

  • Optimize thread pool performance.

• Bug fixes

  • For Linux kernels earlier than 4.9, ppoll is disabled and poll is used instead.

  • The bug that causes errors when the wrap_sm4_encrypt function is invoked is fixed.

  • The bug due to which the system locks global variables during the rotation of audit logs is fixed.

  • The bug that causes errors during restoration inconsistency checks is fixed.

  • The bug that causes inaccurate time values in the io_statistics table is fixed.

  • The bug due to which an RDS instance unexpectedly exits when invalid compression algorithms are used is fixed.

  • The bug that causes user columns in MySQL 8.0 and MySQL 5.6 to be incompatible is fixed.

  • The bug that causes buffer count leaks is fixed.

Minor version

Community version

Description

Expiration date

20251220

5.7.44

• New features

  • A statistical item for CPU time consumption is added to audit logs.

• Bug fixes

  1. Fixed an issue where the Sequence engine caused the instance to restart when recording binary logging and rotating binary logs.

  2. Fixed the long duration during version upgrades when enabling General Query Log.

  3. Fixed performance issues caused by lock wait in trx_sys during high concurrency.

20251031

5.7.44

• New features

  • The length of INT, VARCHAR, and CHAR data types can be extended online.

  • The character set of a column can be instantly changed.

  • The NET::buff memory for persistent connections is automatically released to reduce memory resource usage and lower the risk of OOM errors.

  • The built-in functions statement_digest and statement_digest_text are added.

• Bug fixes

  • The bug that causes replication interruption when the SUPER permission is granted in a native replication scenario is fixed. After the fix, only the permissions owned by the replication thread are granted.

  • The community bug #105652, which causes an exception during crash recovery for large tables that exceed 32 TB, is fixed.

  • The bug that may cause a crash when a DDL statement is executed concurrently with an UPDATE statement on a table with foreign keys is fixed.

  • The bug that causes Seconds_Behind_Master spikes for single-statement transactions is fixed.

  • The bug that causes slow acknowledgments in semi-synchronous replication due to purging of relay logs is fixed.

  • The performance issues caused by clearing the writeset are fixed.

  • The community bug #118476, which prevents creating partitions with certain identifiers, is fixed.

--

20250831

5.7.44

• Bug fixes

  • The bug that causes query exceptions due to the innodb_max_mtr_records_for_tmp_table parameter is fixed.

--

20250430

5.7.44

• New features

  • The Visible Index and Invisible Index features are supported.

  • Records the end time of SQL statements in the start_time field of slow query logs.

• Performance optimizations

  • The execution efficiency of the REPLACE function is improved.

  • If GTID auto-positioning is enabled, gap transactions are automatically skipped when the replication of the secondary instance starts.

• Bug fixes

  • The issue that the value of the start_time field is incorrect in the slow query logs for CALL PROCEDURE is fixed.

--

20250331

5.7.44

• New features

  • Supports the instant column addition feature. For more information, see Instant Column Addition.

• Performance optimizations

  • The following bug is fixed: Writeset-based replication is interrupted in scenarios involving unique indexes on string prefixes.

--

20241231

5.7.44

• New features

  • Statement concurrency control can be implemented based on templates. This way, the execution of SQL statements using the same template is limited.

  • Column-based data encryption supports keys managed by KMS.

  • Column-based data encryption supports the AES 256 algorithm.

--

20241130

5.7.44

• New features

  • OpenSSL is updated to version 3.0.15.

• Performance optimizations

  • The period of time that is required for the automatic remediation of the General Query Log during crash recovery is reduced.

• Bug fixes

  • The bug that the instance becomes unavailable because the buffer pool is used up by internal temporary tables is fixed. For more information, see Community Bug#96236.

  • The bug that the instance may crash when you cancel the scale-in of the buffer pool is fixed.

  • The bug that the transaction hangs and cannot be killed due to the invalid value of trx->in_innodb is fixed. For more information, see Community Bug#99643.

  • The bug due to which binary logs are recorded even if the Optimize Table statement fails to be executed is fixed.

  • The bug that the ciphertext in the EncDB column collides is fixed.

--

20240930

5.7.44

• New features

  • The MySQL 5.7 Basic Edition supports native MySQL replication.

--

20240731

5.7.44

• New features

  • The ignorable log events that are generated by the binlog cache free flush feature can be skipped by the binary log dump thread.

  • The encdb parameter can be dynamically modified.

• Performance optimizations

  • Unnecessary server hooks can be deleted to reduce unnecessary lock contention during DML execution.

  • The duration of a lock that occurs when you execute the TRUNCATE TABLE statement is shortened. If you enable the innodb_rds_drop_ahi_ahead parameter, the adaptive hash indexes created for InnoDB tables are deleted before you truncate the InnoDB tables. This shortens the duration of the metadata lock on the InnoDB tables during the execution of the statement and prevents other SQL statements from being blocked.

  • The period of time that is required for the automatic remediation of the General Query Log during crash recovery is reduced.

• Bug fixes

  • The bug due to which the recovery from a system crash is stuck at 99% is fixed.

  • The bug due to which many SQL statements are stuck in the "Waiting for table flush" state during the execution of the Analyze Table statement is fixed.

  • The bug due to which end_log_pos for the binlog cache free flush feature is inaccurate is fixed.

  • The bug due to which a verification error occurs when data is written to pages during the recovery from a system crash is fixed.

  • The bug due to which memory information is not collected by performance_schema when multiple threads are used to initialize a buffer pool is fixed.

  • Fixed a crash caused by encdb timestamp handling.

  • The bug due to which binary logs are recorded even if the Optimize Table statement fails to be executed is fixed.

  • The bug due to which unnecessary AUTO_INC locks may be obtained during the execution of UPDATE statements is fixed.

--

20240430

5.7.44

• Performance optimizations

  • The duration of a lock that occurs when you execute the DROP TABLE statement is shortened. The innodb_rds_drop_ahi_ahead parameter is added. If you enable the parameter, the adaptive hash indexes created for InnoDB tables are deleted before you delete the InnoDB tables. This shortens the duration of the metadata lock on the InnoDB tables during the execution of the DROP TABLE statement and prevents other SQL statements from being blocked.

  • The Binlog Cache Free Flush feature is supported. During the transaction commit phase, this feature converts the temporary binary log cache files of large transactions into binary log files. This reduces the duration of a global lock on binary logs when large transactions are committed, reduces the I/O pressure on RDS instances, and prevents primary/secondary switchovers triggered when the binary logs cannot be written to the RDS instances for a long period of time or unresponsive I/O operations occur on the RDS instances.

• Bug fixes

  • The bug that the value of the AUTO_INCREMENT parameter of the InnoDB table on the secondary RDS instance is inconsistent with the actual value due to the execution of the REPLACE INTO statement is fixed.

  • Fix the issue where the INSERT ... SELECT statement lets you insert data incompatible with the GEOMETRY type.

  • Fix the issue where EncDB returns disordered data types.

  • Fixed an issue that caused an "error 1305: savepoint does not exist." error when creating a backup using Percona mysqldump with the --lock-for-backup option.

--

20240229

5.7.44

• New features

  • The updates to the features in MySQL 5.7.44 are incorporated.

  • Statement Outline is added to support ps-protocol.

  • OpenSSL is updated to version 3.0.12.

  • Encryption rules and users can be queried using EncDB.

• Performance optimizations

  • The issue that causes second-level replication latency is optimized.

• Bug fixes

  • The bug due to which irrelevant errors are reported after the tablespace is discarded is fixed.

  • The following issue is fixed: After the recycle_bin parameter is disabled, read-only instances still trigger a reclaim.

  • The bug that may cause a crash when querying common tables and the system tables information_schema.PARAMETERS or ROUTINES at the same time is fixed.

  • The bug that causes slow logs to fail to be written due to the loss of slow_log.CSV is fixed.

  • The bug of occasional null pointer access caused by rapid unloading after the query cache plugin is loaded is fixed.

  • The following issue is fixed: When multiple statements are merged into one query, the truncation position of multiple statements recorded in audit log V3 is inaccurate.

  • The following issue is fixed: The drop table force is invalid for tables that have foreign key dependencies.

  • The bug that the more_result field in OK packets is abnormal in multi-statement scenarios is fixed. You can use multiple statements to restore the session status in MaxScale.

  • The following issue is fixed: In semi-synchronous replication, binlog file names are incorrectly compared and semi-synchronous replication fails.

  • The following issue is fixed: When the explicit_defaults_for_timestamp parameter is set, EncDB writes a null value as the timestamp.

--

20240115

5.7.43

• Bug fixes

  • The bug due to which EncDB cannot identify which columns contain sensitive information when table aliases are used in SQL queries is fixed.

  • The bug due to which the system fails when the constructor of the DesensitizationAlgo class in EncDB receives invalid input is fixed.

  • The bug that disk space may be exhausted by the logs generated by EncDB when sensitive rule configuration is empty is fixed.

  • The bug that EncDB returns an error when the calculated size of ciphertext and the actual size of generated ciphertext do not match during the handling of users of the noneAcces type is fixed.

--

20231215

5.7.43

• Bug fixes

  • The bug due to which you cannot connect to the server using TLS 1.0 and TLS 1.1 after you upgrade OpenSSL to OpenSSL 3.0.9 is fixed.

--

20231031

5.7.43

• Bug fixes

  • The bug due to which the EncDB parameter settings do not take effect is fixed.

--

20230831

5.7.43

• New features

  • Incorporates official changes from MySQL 5.7.43.

  • OpenSSL is upgraded to version 3.0.9.

  • The shell built-in functions are used to parse parameters, which accelerates instance startup using mysqld_safe.

  • The Confidential database feature is added to implement end-to-end data encryption and protect data security.

• Bug fixes

  • The bug that an RDS instance is unexpectedly restarted because sql_type for the thread pool is not correctly configured is fixed.

  • The bug due to which a self-looping issue is caused by an incorrect fil_addr pointer during the restoration of the undo table lock is fixed.

  • The bug that audit logs do not record multi-statement queries is fixed.

--

20230531

5.7.42

• New features

  • The updates to the features in MySQL 5.7.41 and 5.7.42 are incorporated.

  • OpenSSL is upgraded to version 1.1.1u.

  • During the recovery from a system crash, the release of the undo tablespace lock is accelerated.

  • The resizing of online buffer pools can be cancelled to save time.

• Performance optimizations

  • During the instance startup, multiple threads are used to initialize the buffer pool instances at a time to speed up the startup.

  • During the recovery from a system crash, the rollback method of in-doubt transactions is changed from synchronous rollback to asynchronous rollback to improve the RTO.

  • The initialization performance of buffer pools is optimized. This accelerates the instance startup.

• Bug fixes

  • The bug that initialization fails because the subscript of a parameter array is out of range when multiple threads are used to initialize the buffer pool instances is fixed.

--

20230228

5.7.40

• Bug fixes

  • The bug due to which the memory usage of your RDS instance may gradually increase in specific scenarios after you enable recycle_scheduler is fixed.

  • The bug due to which the purge_table operation cannot be executed because the recycle_scheduler thread in the background conflicts with the active purge_table operation is fixed.

--

20221231

5.7.40

• New features

  • The updates to the features in MySQL 5.7.40 are incorporated.

  • OpenSSL is updated to version 1.1.1s.

• Performance optimizations

  • The AHI feature is optimized:

    • If you disable this feature, memory that is occupied by AHIs is released.

    • The update speed of AHIs is improved.

  • If a transaction modifies a table that does not have a primary key or contain a non-null unique key, a non-null key with auto-increment fields is preferentially used to search for the target column when the modification is applied to the secondary RDS instance. This improves table playback efficiency and reduces replication delay in the secondary RDS instance.

    Note

    A non-null key with auto-increment fields can be a user-defined key or an implicit primary key created by the system.

• Bug fixes

  • The bug due to which the number of rows updated by the UPDATE and INSERT statements in the audit logs of an ApsaraDB RDS for MySQL instance is inaccurate is fixed.

  • The following bug is fixed: When a temporary table is used to create a standard table, the 13146 error is reported for the secondary RDS instance because the character set of the temporary table is missing in binary logs. After the bug is fixed, the character set of the temporary table is recorded in binary logs when a temporary table is used to create a standard table. The 13146 error is no longer reported for the secondary RDS instance.

--

20221031

5.7.39

• New features

  The updates to the features in MySQL 5.7.39 are incorporated.

• Bug fixes

  The primary fast look up feature is removed.

--

20220731

5.7.38

• New features

  The updates to the features in MySQL 5.7.38 are incorporated.

• Bug fixes

  • The bug that causes deadlocks due to the conflict between Rename Tablespace operations required in DDL operations and ongoing I/O operations is fixed.

  • The bug that causes inaccurate statistics on the Memory/sql/String::value event in the memory_summary_global_by_event_name table of the performance_schema database is fixed.

  • The bug that causes deadlocks for XA transactions on the Replica node due to improper locking of secondary indexes when data is inserted into secondary indexes is fixed.

  • Removes the restriction that innodb_buffer_pool_size must be at least 1 GB when innodb_buffer_pool_instances exceeds 1.

• Performance optimizations

  The policy to adjust the size of a buffer pool is optimized to make the adjustment process smoother.

--

20220430

5.7.37

• New features

  • The updates to the features in MySQL 5.7.37 are incorporated.

  • The SSL encryption feature is optimized, and a few bugs of the feature are fixed.

  • The permissions to manage the sys or performance_schema database are revoked from standard accounts.

• Bug fixes

  The level of the FLUSH TABLES WITH READ LOCK and UNLOCK TABLES logs in error logs is changed to warning. This fix facilitates online troubleshooting.

--

20211231

5.7.35

• New features

  • The updates to the features in MySQL 5.7.35 are incorporated.

  • The InnoDB buffer pool can be flexibly resized. The impact of a resizing operation is reduced, and user experience is improved.

  • The bidirectional replication mechanism is optimized. Unnecessary binary logs are no longer replicated. In addition, heartbeat events are no longer generated when events are skipped.

  • You no longer need to restart your RDS instance after you reconfigure the log_slave_updates parameter.

  • FLUSH TABLES WITH READ LOCK and UNLOCK TABLES are added to the error log.

  • The innodb_control_index_page_reserve parameter is added. If this parameter is set to OFF, page splitting and storage usage are reduced.

• Bug fixes

  • The bug due to which the performance agent feature cannot collect CPU usage statistics as expected in the new docker architecture is fixed.

  • The bug due to which session tracker type numbers retained in the system are incorrect is fixed.

  • The bug due to which an RDS instance unexpectedly exits and cannot be restored to normal when the size of mtr logs exceeds the size of the log parsing buffer is fixed.

  • Memory pages can be marked as MADV_DONTDUMP. If the innodb_buffer_pool_in_core_file parameter is disabled, core dumps no longer contain the InnoDB buffer pool. This way, the storage that is occupied by core dumps is reduced.

  • The bug due to which the value of the start_time field in the mysql.slow_log table is incorrect is fixed. After the bug is fixed, the field records the time at which the execution of an SQL statement starts rather than the time at which the execution of an SQL statement ends.

  • The Show_db_priv permission is available to standard accounts.

  • The bug due to which memory leaks may occur when the LF_HASH_INSERT function is invoked is fixed.

  • The bug due to which the recycle_bin does not support tables that contain virtual columns is fixed.

--

20210630

5.7.32

• New features

  • The audit log format MYSQL_V3 is supported.

  • Slow query logs and general logs can be truncated.

  • A mechanism is provided to detect out of memory (OOM) exceptions in the thread stack.

  • The thread_pool_strict_mode parameter is supported. This parameter is used to limit the maximum number of worker threads.

• Bug fixes

  • When starting mysqld, the mysqld_safe script deletes old socket lock files.

  • The bug that causes memory leaks in the recycle_bin is fixed.

--

20210430

5.7.32

• New features

  • The recycle bin feature is supported. For more information, see Recycle Bin.

  • The Sequence engine feature is supported. For more information, see Sequence Engine.

• Performance optimizations

  • More metrics that are used to measure the performance of an RDS instance are supported for the I_S.PERF_STATISTICS table.

  • The metric that specifies the number of index scans is supported for the I_S.INDEX_STATISTICS table.

  • The TDE performance is optimized.

• Bug fixes

  The bug that causes column generation exceptions during a rollback is fixed. This bug is found in MySQL Community Edition.

--

20201031

5.7.30

• Bug fixes

  • The bug due to which the ROW_SEARCH_MVCC function unexpectedly exits when concurrent updates are performed is fixed.

  • The bug due to which an RDS instance cannot start when the innodb_undo_tablespaces parameter is reconfigured for the RDS instance is fixed.

  • The bug that causes a cache overflow when FTS-based queries are run is fixed.

20240601

20200831

5.7.30

• New features

  • The updates to the features in open source MySQL 5.7.30 are incorporated.

  • Optimizes CCL's waiting and concurrent mechanisms.

  • Start GTID (start gtid) and end GTID (stop gtid) are supported by mysqlbinlog.

  • LSNs are supported in the redo log:

    • innodb_lsn: The LSN of the redo log.

    • innodb_log_write_lsn: The Log Sequence Number (LSN) for log writes.

    • innodb_log_checkpoint_lsn: the LSN of the last checkpoint.

    • innodb_log_flushed_lsn: The LSN of the last redo log record flushed to disk.

    • innodb_log_pages_flushed: The log sequence number (LSN) of the flushed pages.

• Performance optimizations

  • Adjust the execution priority of concurrency control in a stored procedure.

• Bug fixes

  • During SQL execution, pages allocated for temporary tables may experience a reference count leak. This can cause inefficient flushing of dirty pages in the buffer pool, leading the buffer pool to run out of free pages and severely degrading database performance. For more information, see the official documentation.

20240601

20200630

5.7.28

• New features

  • Inventory Hint: Adds three new hints to support SELECT, UPDATE, INSERT, and DELETE statements, quickly commit or roll back transactions, and increase application throughput capacity.

  • SQL throttling: Controls concurrency to handle sudden spikes in database request traffic, access to statements that consume excessive resources, and changes in SQL access patterns, ensuring continuous and stable operation of MySQL instances.

  • Statement Queue: The Statement Queue is a queuing mechanism that uses bucket-based queuing to place statements that are likely to conflict with each other into the same bucket, reducing the overhead caused by conflicts.

  • Statement Outline: Use Optimizer Hints and Index Hints to stabilize MySQL's execution plan.

  • Faster DDL: Optimizes the Buffer Pool management mechanism during DDL operations to reduce the performance impact of DDL operations and increase concurrency for online DDL operations.

  • The maximum number of connections that are allowed is increased to 500,000.

• Performance optimizations

  • You can view all native procedures by running the call dbms_admin.show_native_procedure(); command.

  • A new function is provided to delete orphan tables.

  • The thread pool feature is optimized.

  • Query caching is optimized.

  • You can set the maximum memory for Performance Schema according to the instance type.

• Bug fixes

  • The bug due to which an audit update thread enters an infinite loop is fixed.

20240601

20200430

5.7.28

• Performance optimizations

  • Read-write locks are no longer supported in the query cache. The default hash function is changed from LF_hash to murmur3 hash.

• Bug fixes

  • Two bugs that occur when the system hits the requested data in the query cache are fixed. These bugs are triggered if transactions at the REPEATABLE_READ isolation level are run.

20240601

20200331

5.7.28

• New features

  • Merge the official changes from MySQL 5.7.28.

  • Fast Query Cache: Addresses the limitations of the native MySQL Query Cache by redesigning and implementing it from scratch, launching the RDS Query Cache to significantly improve database query performance.

  • Two metadata locks from percona-server 5.7 are supported: LOCK TABLES FOR BACKUP (LTFB) and LOCK BINLOG FOR BACKUP (LBFB).

• Performance optimizations

  • Added backward compatibility for thread pools.

  • The output of TCP errors is disabled by default.

  • The thread pool feature is optimized to provide higher performance when the default configurations are used.

• Bug fixes

  • The bug due to which the system deletes temporary files when large files are deleted is fixed.

  • The bug due to which dump threads in thread pools time out is fixed.

  • The bug due to which the system incorrectly counts the value of the IPK field in the procedure context is fixed.

  • The bug that causes the leak and release of pfs threads when the rds_change_user command is run is fixed.

20240601

20200229

5.7.26

• New features

  • Performance Agent: A MySQL plugin that collects and analyzes internal performance data from MySQL instances.

  • In semi-synchronous mode, record the network round trip time in the performance data.

• Performance optimizations

  • The amount of time that is required to execute the pause instruction is reduced in various CPU architectures.

  • The database proxy feature is enhanced to optimize short-lived connections.

  • A memory summary table is added to provide the statuses of thread pools.

• Bug fixes

  • The bug that compromises the security of DDL-related redo log records is fixed.

  • The bug that causes inaccurate time values in the io_statistics table is fixed.

  • The bug due to which an RDS instance unexpectedly exits when tables are modified is fixed.

  • The bugs in MySQL test cases are fixed.

20240601

20200110

5.7.26

• Performance optimizations

  • Unlink small files when purging files asynchronously.

  • You can optimize the performance of the Thread Pool.

  • The default value of the thread_pool_enabled parameter is changed to OFF.

20240601

20191225

5.7.26

• New features

  • The management of internal accounts is supported. This lets you manage permissions and protect data in a more efficient manner.

• Performance optimizations

  • The mechanism that is used to process short-lived connections is optimized.

  • A dedicated thread is used for the maintain user service. This prevents HA failures.

  • Unnecessary TCP error logs can be deleted.

  • The thread pool feature is optimized.

• Bug fixes

  • The bug due to which the mysqld process unexpectedly exits when the read/write splitting feature is enabled is fixed.

  • The bug that causes errors in core dumps when a keyring is used is fixed.

20240601

20191115

5.7.26

• Bug fixes

  • The bug due to which the system displays variables in audit logs when a primary/secondary switchover is performed is fixed.

20240601

20191101

5.7.26

• New features

  • Add support for the SM4 encryption algorithm to TDE.

  • If you specify a primary key, the system directly accesses the primary index.

  • A mechanism is provided to prevent the automatic conversion of tables from the Memory engine to the MyISAM engine. These tables include system tables and the tables that are called by threads in the initializing state.

• Performance optimizations

  • Thread Pool is optimized to reduce mutexes.

  • An audit log caching mechanism is provided to improve logging performance.

  • Performance Insight: Performance points support thread pools.

  • The thread pool feature is enabled by default. Thread Pool

• Bug fixes

  • You can release the lock when processing the maintenance user list.

  • More TCP errors are added.

20240601

20191015

5.7.26

• New features

  • The rotation of slow query logs is supported. Each CSV slow query log file is assigned a unique name and saved as a new file. This prevents data loss during the collection of slow query logs. You can run the show variables like '%rotate_log_table%'; command to check whether the rotation of slow query logs is enabled.

  • A performance agent plugin is provided. This plugin obtains performance data and saves the data as text files to on-premises computers. A round-robin algorithm is used to retain only the most recent files that contain performance data at the single-digit second level.

  • Forced conversion from the MEMORY storage engine to the InnoDB engine is supported. If the global variable rds_force_memory_to_innodb is set to ON, tables are automatically converted from the MEMORY storage engine to the InnoDB engine when the tables are created or modified.

  • The TDE mechanism is optimized: The keyring-rds plugin is added to interact with the control system/Key Management Service.

  • The output of TCP errors is supported. TCP read, read-wait, and write-wait errors are returned together with error codes in end_connection events. Log records that contain the information about the errors are also generated.

• Bug fixes

  • The bug that causes error 1290 in DDL operations is fixed.

20240601

20190925

5.7.26

• Parameter changes

  • The default value of the system variable auto_generate_certs is changed from true to false.

  • The global read-only variable auto_detact_certs is added. Default value: false. Valid values: [true | false]. This system variable is available when the server side is compiled with OpenSSL. It controls whether the server automatically searches for SSL certificate and key files in the data directory at startup, which means it controls whether to enable the automatic search for certificates and keys on the server side.

20240601

20190915

5.7.26

• New features

  • Thread Pool: Separates threads from sessions, which allows many sessions to be handled by a small number of threads that complete tasks for active sessions.

20240601

20190815

5.7.26

• New features

  • Purge Large File Asynchronously: When you delete a single tablespace, the system renames the tablespace files as temporary files and the asynchronous purge process cleans up the temporary files.

  • Performance Insight: Focuses on instance load monitoring, association analysis, and performance tuning to help you quickly evaluate database loads, find the source of performance issues, and improve database stability.

  • An optimized instance locking mechanism is provided. This mechanism lets you drop or truncate tables on an RDS instance even if the RDS instance is locked.

• Bug fixes

  • Do not set rds_prepare_begin_id in the set rds_current_connection command.

  • You can modify the information for users marked as (locked).

  • Do not use actual as a table name.

  • The bug that causes an overflow of timestamps in slow query logs is fixed.

20240601

20190510

5.7.26

• New features

  • The creation of temporary tables in transactions is supported.

20240601

20190319

5.7.26

• New features

  • The configuration of thread IDs for proxies in handshake packets is supported.

20240601

20190131

5.7.25

• Performance optimizations

  • Upgrade to the official version 5.7.25.

  • JeMalloc that is used for memory management is disabled.

• Bug fixes

  • The bug due to which the system incorrectly calculates the value of the internal variable net_lenth_size is fixed.

20240601

20181226

--

• New features

  • Dynamic modifications to the system variable binlog-row-event-max-size are supported. This accelerates the replication of tables that do not have a primary key.

• Bug fixes

  • The bug due to which the proxy instances of an RDS instance cannot apply for memory resources is fixed.

20240601

20181010

--

• Performance optimizations

  • Implicit primary keys are supported.

  • The replication of tables that do not have a primary key between a primary RDS instance and its secondary RDS instance is accelerated.

  • Native AIO is provided to improve I/O performance.

20240601

20180431

--

• New features

  • It supports high-availability series.

  • The SQL Audit feature is supported. For more information, see SQL Audit.

  • The protection for RDS instances on which snapshot backups are being created is enhanced.

20240601

Minor version

Description

20230101

• Bug fixes

  • When multiple sessions insert into a table that has an auto-increment column, IODKU is used as the unique key constraint, and an explicit value for the auto-increment column is not added to the table. As a result, the insert operation may fail due to a unique index conflict. After the fix, the auto-increment ID of the row into which IODKU fails to insert data is retained for the next insert operation, and this failed insert operation is treated as an update operation.

20201229

• Bug fixes

  • The bug that causes unexpected database crashes and hangs is fixed.

  • The bug that causes prefix indexes to be incorrectly identified as invisible indexes is fixed.

  • The bug that causes abnormal states in logger nodes based on the consensus protocol is fixed. If this bug persists, ApsaraDB RDS may fail to elect an eligible RDS instance as the primary RDS instance.

20191128

• New features

  • The read/write splitting feature is supported.

• Bug fixes

  • The bug that causes incorrect calculation of the Second_Behind_Master metric for followers in some scenarios is fixed.

  • The bug that causes deadlocks when you retry table-level parallel replication transactions is fixed.

  • XA-related bugs are fixed.

20191016

• New features

  • RDS instances that run MySQL 5.7 on RDS High-availability Edition (Premium Local SSD) can be upgraded to RDS Enterprise Edition.

  • MySQL official GTID functionality is supported. GTIDs are disabled by default.

  • All proprietary AliSQL features released for RDS Basic Edition and RDS High-availability Edition in versions up to 20190915 are incorporated.

• Bug fixes

  • The bug that disables binary logs for a primary RDS instance after its secondary RDS instance is reset is fixed.

20190909

• New features

  • The execution efficiency of large transactions in three-node strong-consistency mode is optimized.

  • Binary log records can be dumped from leader or follower nodes.

  • Read-only RDS instances can be created.

  • System tables use the InnoDB engine by default.

• Bug fixes

  • The bug that invalidates the log cleanup commands for followers is fixed.

  • The bug that causes slave threads to exit unexpectedly is fixed. This bug occurs when the slave_sql_verify_checksum parameter is set to OFF and the binlog_checksum parameter is set to crc32.

20190709

• New features

  • It supports the three-node feature.

  • The semi-sync plugin is disabled.

  • Table-level parallel replication and writeset-based parallel replication are supported.

  • The pk_access module is provided to accelerate queries that are run based on primary keys.

  • The thread pool feature is supported.

  • All proprietary AliSQL features released for RDS Basic Edition and RDS High-availability Edition in versions up to 20190510 are incorporated.

Minor version

Community version

Description

Expiration date

20250531

5.6.16

• Bug fixes

  • Fixed an offset comparison error in semi-sync that caused a memory leak in the primary database.

--

20241231

5.6.16

• Bug fixes

  • The bug that causes unexpected restarts of thread pool-enabled instances due to sql_type errors is fixed.

  • The bug that causes replication interruption due to incorrect comparison of binary log file names in semi-synchronous replication is fixed.

--

20221130

5.6.16

• New features

  Information about GLOBAL READ LOCK acquisition and release is recorded in error logs.

• Bug fixes

  • The bug that causes excessive memory usage when many InnoDB partitioned tables are opened is fixed.

  • The bug that causes the InnoDB data dictionary to remain unchanged when column names change case is fixed.

  • The bug that causes MySQL instances to deadlock when tablespaces are renamed is fixed.

  • The bug that causes crash recovery failures due to a small recovery parsing buffer is fixed.

  • The bug that causes inaccurate results for the SHOW GLOBAL STATUS; statement is fixed.

  • The bug that causes inaccurate timestamps in the start_time field of the mysql.slow_log table is fixed.

--

20221126

5.6.16

• Bug fixes

  • The bug that causes duplicate entries in the AUTO_INCREMENT column of InnoDB tables is fixed.

--

20210630

5.6.16

• New features

  A mechanism is provided to detect out-of-memory (OOM) exceptions in the thread stack.

• Bug fixes

  • The bugs in MySQL 5.6 Community Edition are fixed.

  • The operation limits on the SHOW CREATE statement in the mysql system database are removed.

--

20210430

5.6.16

• Performance optimizations

  More metrics that are used to measure the performance of an RDS instance are supported for the I_S.PERF_STATISTICS table.

• Bug fixes

  The bugs that cause memory leaks in MySQL Community Edition are fixed.

--

20201031

5.6.16

• Bug fixes

  • The bug that invalidates subqueries in IN clauses is fixed.

  • The bug that causes process permission errors is fixed.

  • The bug that causes permission issues for users in the kill_user_list table is fixed.

  • The bug that causes errors during the execution of the DROP DATABASE statement is fixed.

  • The bug that causes incorrect SECONDS_BEHIND_MASTER calculations due to PREVIOUS_GTID events is fixed.

20240601

20200831

5.6.16

• New features

  LSNs are supported in the redo log:

  • innodb_lsn: The LSN of the redo log.

  • innodb_log_write_lsn: The log sequence number (LSN) for log writes.

  • innodb_log_checkpoint_lsn: The LSN at which the last checkpoint occurred.

  • innodb_log_flushed_lsn: The log sequence number (LSN) of the redo log that has been flushed to disk.

  • innodb_log_pages_flushed: the log sequence number (LSN) of flushed pages.

• Bug fixes

  • The bug that causes type errors in the SHOW_HA_ROWS statement is fixed.

  • The bug that causes incorrect counting of the IPK field in procedure contexts is fixed.

  • The bug that causes server crashes when INFORMATION_SCHEMA tables are queried is fixed.

  • The bug that causes audit update threads to enter infinite loops is fixed.

  • The bug that prevents secondary RDS instances from reporting replication delays is fixed.

20240601

20200630

5.6.16

• New features

  • Performance Agent: This feature provides a more convenient way to collect and analyze performance data. It is provided as a MySQL extension and allows the system to compute and analyze the metrics that are used to measure the performance of an RDS instance.

  • The maximum number of connections that are allowed is increased to 500,000.

  • Faster DDL: Optimizes the Buffer Pool management mechanism for DDL operations to reduce their performance impact and increase the concurrency of online DDL operations.

• Performance optimizations

  • The global max_execution_time parameter is added. When the execution time of an SQL statement exceeds this value, the SQL statement is interrupted.

  • The thread pool feature is optimized.

• Bug fixes

  • The bug due to which the system incorrectly counts the number of waits when commands are read from a database client is fixed.

  • The bug that prevents standard accounts from executing the DROP DATABASE command is fixed.

20240601

20200430

5.6.16

• New features

  • The mdl_info table is added to store metadata lock information.

• Bug fixes

  • The bug that causes conflicts between the thread pool and ic_reduce (flash sale) features is fixed.

20240601

20200331

5.6.16

• Performance optimizations

  • The thread pool feature is optimized to provide higher performance when the default configurations are used.

  • The output of TCP errors is disabled by default.

• Bug fixes

  • The bug due to which temporary files are deleted when large files are deleted is fixed.

20240601

20200229

5.6.16

• New features

  • The proxy read/write splitting feature is supported.

• Performance optimizations

  • The thread pool feature is optimized.

  • The amount of time that is required to execute the PAUSE instruction is reduced in various CPU architectures.

• Bug fixes

  • The bug that causes partial commits of XA transactions is fixed.

20240601

20200110

5.6.16

• New features

  • Thread Pool: This feature separates threads from sessions, allowing you to handle tasks for active sessions with only a small number of threads, even when you have many sessions.

• Performance optimizations

  • During asynchronous file purging, cancel links to small files.

• Bug fixes

  • The bug due to which the sleep time of the page cleaner is incorrectly calculated is fixed.

  • The bug that causes failover failures when the SELECT @@global.gtid_executed statement is executed is fixed.

  • The bug that causes the "IF CLIENT KILLED AFTER ROLLBACK TO SAVEPOINT PREVIOUS STMTS COMMITTED" error is fixed. For more information, see IF CLIENT KILLED AFTER ROLLBACK TO SAVEPOINT PREVIOUS STMTS COMMITTED.

20240601

20191212

5.6.16

• Performance optimizations

  • Unnecessary TCP error logs are deleted.

20240601

20191115

5.6.16

• Bug fixes

  • The bug that causes timestamp overflows in slow query logs is fixed.

20240601

20191101

5.6.16

• Bug fixes

  • The bug that causes slow query logs to rotate when logs are updated is fixed. Slow query logs are rotated only when the slow query log rotation command is executed.

  • Fixed some display errors.

20240601

20191015

5.6.16

• New features

  • The rotation of slow query logs is supported. Each CSV slow query log file is assigned a unique name and saved as a new file. This prevents data loss during the collection of slow query logs. You can run the show variables like '%rotate_log_table%'; command to check whether the rotation of slow query logs is enabled.

  • The SM4 encryption algorithm is added to replace the outdated SM encryption algorithm.

  • Purge Large File Asynchronously: When you delete a single tablespace, the system renames the tablespace files as temporary files and waits for the asynchronous purge process to clean up the temporary files.

  • The output of TCP errors is supported. TCP read, read-wait, and write-wait errors are returned together with error codes in end_connection events. Log records that contain the information about the errors are also generated.

  • An audit log caching mechanism is provided to improve logging performance.

• Bug fixes

  • The pstack command is disabled to prevent unresponsive behavior when many connections exist.

  • The bug that causes conflicts between implicit primary keys and the create table as select statement is fixed.

  • Temporary files created from binary logs are automatically cleared.

20240601

20190815

5.6.16

• Performance optimizations

  • An optimized instance locking mechanism is provided. This mechanism lets you drop or truncate tables on an RDS instance even if the RDS instance is locked.

20240601

20190130

5.6.16

• Bug fixes

  • Fixed several bugs that could cause system instability.

20240601

20181010

5.6.16

• Performance optimizations

  • The rocksdb_ddl_commit_in_the_middle parameter is supported for MyRocks. If this parameter is enabled, some DDL statements perform commit operations during execution.

20240601

201806**

5.6.16

• New features

  • The precision of slow query logs is improved to microseconds.

20240601

20180426

5.6.16

• New features

  • Invisible indexes are supported. For more information, see Reference documentation.

• Bug fixes

  • The bug that affects apply threads on secondary RDS instances is fixed.

  • The bug that compromises database performance when updates to partitioned tables are applied on secondary RDS instances is fixed.

  • The bug that causes TokuDB to rebuild entire tables when the ALTER TABLE COMMENT statement is executed on these tables is fixed. For more information, see Reference documentation.

  • The bug that causes deadlocks when the SHOW SLAVE STATUS or SHOW STATUS statement is executed is fixed.

20240601

20171205

5.6.16

• Bug fixes

  • The bug that causes deadlocks when OPTIMIZE TABLE and ONLINE ALTER TABLE statements are executed simultaneously is fixed.

  • The bug that causes conflicts between sequences and implicit primary keys is fixed.

  • The bug that causes the SHOW CREATE SEQUENCE statement to fail is fixed.

  • The bug that causes inaccurate statistics for TokuDB tables is fixed.

  • The bug that causes deadlocks when OPTIMIZE statements are executed on multiple tables in parallel is fixed.

  • The bug that causes inaccurate character sets in QUERY_LOG_EVENT records is fixed.

  • The bug that prevents databases from stopping correctly due to signal handling errors is fixed. For more information, see Reference documentation.

  • The bug that causes RESET MASTER to fail is fixed.

  • The bug that causes secondary RDS instances to hang is fixed.

  • Resolved an issue with status maintenance after a primary node failover in RDS Enterprise Edition.

  • The bug that causes processes to crash unexpectedly when the SHOW CREATE TABLE statement is executed is fixed.

20240601

20170927

5.6.16

• Bug fixes

  • The bug that causes TokuDB tables to use incorrect indexes during queries is fixed.

20240601

20170901

5.6.16

• New features

  • The SSL encryption version is upgraded to TLS 1.2. For more information, see Reference documentation.

  • Sequences are supported.

• Bug fixes

  • The bug that causes incorrect result sets for NOT IN queries in specific scenarios is fixed.

20240601

20170530

5.6.16

• New features

  • Privileged accounts can kill connections established by standard accounts.

20240601

20170221

5.6.16

• New features

  • Supports enabling the agent terminal.

20240601

Minor version

Description

20181212

• Bug fixes

  • The bug that causes inaccurate return values for the gettimeofday(2) system function is fixed. This system function returns time values, which are commonly used to calculate wait timeouts. Inaccurate time values prevent correct timeout calculations.