Aurora MySQL database engine updates 2024-03-07 (version 3.06.0, compatible with MySQL 8.0.34)

Aurora MySQL version 3.06.0 supports Amazon Bedrock integration and introduces the new reserved keywords accept, aws_bedrock_invoke_model, aws_sagemaker_invoke_endpoint, content_type, and timeout_ms. Check the object definitions for the usage of the new reserved keywords before upgrading to version 3.06.0. To mitigate the conflict with the new reserved keywords, quote the reserved keywords used in the object definitions. For more information on the Amazon Bedrock integration and handling the reserved keywords, see What is Amazon Bedrock? in the Amazon Aurora User Guide. For additional information, see Keywords and Reserved Words, The INFORMATION_SCHEMA KEYWORDS Table, and Schema Object Names in the MySQL documentation.

Improved performance for binary log replicas when replicating transactions for large tables with more than one secondary index. This feature introduces a thread pool to apply secondary index changes in parallel on a binlog replica. The feature is controlled by the aurora_binlog_replication_sec_index_parallel_workers DB cluster parameter, which controls the total number of parallel threads available to apply the secondary index changes. For more information, see Optimizing binary log replication in the Amazon Aurora User Guide.

Added a new stored procedure mysql.rds_set_read_only that allows changing the value of the global system variable read_only on database instances in your Aurora MySQL cluster. For more information, see Replicating in the Amazon Aurora User Guide.

Added a new stored procedure mysql.rds_set_binlog_source_ssl that allows setting the encryption on a binary log replica by specifying a value for SOURCE_SSL. For more information, see Replicating in the Amazon Aurora User Guide.

Amazon Aurora Machine Learning is an optimized integration between the Aurora MySQL database and AWS machine learning (ML) services. Amazon Bedrock is now supported, allowing you to invoke machine learning models in Amazon Bedrock directly from your Aurora MySQL DB cluster using SQL. For more information on using Amazon Bedrock with your Aurora MySQL DB cluster, see Using Amazon Aurora machine learning with Aurora MySQL in the Amazon Aurora User Guide.

Aurora MySQL version 3.06 adds support for automated undo tablespace truncation. This optimization allows you to reclaim unused space in undo tablespaces after the undo logs have been purged.

CVE-2020-11104

CVE-2020-11105

CVE-2023-38545

CVE-2023-38546

CVE-2023-39975

Fixed an issue where a read replica DB instance can't be launched successfully when there's high workload in the writer DB instance.

Fixed an issue where an Aurora MySQL writer DB instance can fail over due to a defect in communication with Aurora storage. The defect occurs as a result of a breakdown in the communication between the DB instance and underlying storage following a software update of the Aurora storage instance.

Fixed an issue when processing INSERT queries on InnoDB partitioned tables that can cause a gradual decline of free memory in the instance.

Fixed an issue that can cause an Aurora MySQL DB instance to restart or fail over due to a decrease in freeable memory when hash join is used while running queries.

Fixed an issue that can cause a database instance restart when running SHOW STATUS and PURGE BINARY LOGS statements concurrently. PURGE BINARY LOGS is a managed statement that is run to honor the user-configured binlog retention period.

Fixed an issue that can cause the server to unexpectedly close after running Data Manipulation Language (DML) statements on a table whose nonvirtual columns were reordered with a MODIFY COLUMN or CHANGE COLUMN statement.

Fixed an issue that, during the restart of a database instance, can cause an additional restart.

Fixed an issue that can cause a database restart when a cascading UPDATE or DELETE foreign key constraint is defined on a table where a virtual column is involved either as a column in the foreign key constraint, or as a member of the referenced table.

In Aurora MySQL 2.10, we added support for rebooting an Aurora DB cluster with read availability. This feature allows reader DB instances to stay online while a writer DB instance is rebooted. This feature is now supported on secondary AWS Regions in Aurora MySQL global databases, ensuring that you can still serve read requests during a writer instance restart on the primary cluster. Previously, when a writer instance restarted, all reader instances in an Aurora MySQL secondary cluster also restarted. With this release, secondary cluster reader instances continue to serve read requests during a writer instance restart, improving read availability in the cluster. For more information, see Rebooting an Aurora cluster with read availability.

Fixed an issue that can interrupt database recovery during startup if the restart occurred while running heavy insert operations involving AUTO_INCREMENT columns.

Fixed an issue that can cause a parallel query to fail due to transient network issues while reading data from the Aurora cluster volume.

Fixed an issue where the user is unable to interrupt any query or set session timeouts for performance_schema queries.

Fixed an issue where binary log (binlog) replication configured to use custom SSL certificates (mysql.rds_import_binlog_ssl_material) could fail when the replication instance is undergoing a host replacement.

Small DB instances with less than or equal to 4 GiB of memory now close the top memory-consuming connections when the DB instance is under memory pressure. You can also tune the buffer pool to decrease its size. For more information, see Amazon Aurora MySQL out-of-memory issues in the Amazon Aurora User Guide.

Changed the default response for aurora_oom_response, on all DB instance classes that have more than 4 GiB of memory, from empty to print. For more information, see Amazon Aurora MySQL out-of-memory issues in the Amazon Aurora User Guide.

Fixed an issue related to audit log file management that can cause log files to be inaccessible for download or rotation, and in some cases increase CPU usage.

Optimized AUTO_INCREMENT key recovery to reduce the completion time for restoring snapshots, performing point-in-time recovery, and cloning DB clusters with large numbers of tables in the database.

Fixed an issue where the wait/io/redo_log_flush event wasn't shown in the Performance Schema wait event summary tables.

Added the Aurora_lockmgr_memory_used and Aurora_lockmgr_buffer_pool_memory_used metrics to track the lock manager's memory usage. For more information, see Aurora MySQL global status variables in the Amazon Aurora User Guide.

Fixed an issue where small read replica instances can experience increased replication lag after upgrading from Aurora MySQL versions lower than 2.11.*.

Fixed an issue that can cause duplicate key errors for AUTO_INCREMENT columns using descending indices after a snapshot restore, backtrack, or database cloning operation.

Fixed an issue where a SELECT query on an Aurora reader instance might fail with the error table doesn't exist when the table has at least one full-text search (FTS) index and a TRUNCATE statement is being run on the Aurora writer DB instance.

Fixed an issue that can cause an incomplete result set when running queries involving LEFT JOIN or RIGHT JOIN operations using the hash join algorithm with parallel query.

Fixed an issue that can cause major version upgrades to fail if there is a user-defined FTS_DOC_ID column present in the table schema.

Fixed an issue that can cause upgrade failures from Aurora MySQLversion 2 to Aurora MySQL version 3 due to a synchronization issue while processing InnoDB tablespaces.

Fixed an issue that can cause major version upgrades to Aurora MySQL version 3 to fail due to the presence of orphaned entries for already deleted tablespaces in InnoDB system tables in Aurora MySQL version 2.

Fixed an issue where cache line value can be calculated incorrectly, causing a failure during database restart on Graviton-based instances. (Community Bug Fix #35479763)

Fixed an issue where some instances of subqueries within stored routines were not always handled correctly. (Community Bug Fix #35377192)

Fixed an issue that can cause higher CPU usage due to background TLS certificate rotation (Community Bug Fix #34284186).

Fixed an issue where InnoDB allowed the addition of INSTANT columns to tables in the MySQL system schema in Aurora MySQL versions lower than 3.05, which could lead to the server unexpectedly closing (database instance restarting) after upgrading to Aurora MySQL version 3.05.0. (Community Bug Fix #35625510).