Aurora MySQL database engine updates 2025-05-05 (version 3.04.4, compatible with MySQL 8.0.28)

CVE-2024-11053

CVE-2024-37371

Fixed an issue on the replica where a network interruption may not correctly re-establish connection with the writer.

Fixed an issue that can cause a restart on a binary log (binlog) replica when processing a large number of relay log files during relay log recovery.

Fixed an issue that causes a database reader instance restart when executing a query using the Parallel Query feature.

Fixed an issue that, in rare conditions, can disable binary logging when an error occurs during commit of a large transaction.

Fixed an issue that can cause Aurora read replicas to restart in the event of certain rare transaction commit orders on the writer DB instance.

Fixed an issue that can lead to a database restart when scheduled events are aborted during execution on instances that have Enhanced Binlog enabled.

Fixed an issue where database instances using multi TB Aurora cluster volumes, may experience increased downtime during restart due to InnoDB bufferpool validation failures.

The following privileges have been added to the rds_superuser_role: FLUSH_OPTIMIZER_COSTS, FLUSH_STATUS, FLUSH_TABLES, and FLUSH_USER_RESOURCES. For information about the rds_superuser_role, see Amazon Master User Accounts with Amazon Aurora in the Amazon Aurora User Guide. For more information on these dynamic privileges, please see the MySQL documentation.

Fixed an issue where a preserved connection is handled incorrectly during Zero-downtime patching (ZDP)/Zero-downtime restart (ZDR) that can lead to the client waiting indefinitely for a query to complete.

Fixed an issue where the row becomes unreadable through the spatial index during an update.

Fixed an issue where a query containing an optimizer hint that was aborted during a Zero-downtime restart or Zero-downtime patching operation may be incorrectly handled.

Fixed an issue where the commit latency is not measured when innodb_flush_log_at_trx_commit is set to 0.

Introduced optimizations to reduce memory usage during logical data dictionary recovery when there are a large number of tables.

Fixed an issue that caused the SHOW BINARY LOGS command to take longer to execute on a cluster where Enhanced Binlog is enabled or was previously enabled. This issue could also cause increased commit latency if multiple SHOW BINARY LOGS commands were running concurrently.

Fixed a memory issue associated with the default roles of the view definer.

Fixed an issue that can cause failure in completing the process of disabling the write forwarding feature.

Fixed an issue that can prevent new client connections from being established to the database when write forwarding is enabled.

Fixed an issue that can cause a writer database instance to restart when a reader instance using write forwarding executes a DML statement that contains a timestamp value and the time_zone database parameter is set to "UTC".

Fixed an issue that caused intermittent unavailability of an Aurora Read Replica or table definition inconsistencies with error 'Table does not exist' on the replica due to concurrent read queries on the replica and DDL operations on the writer.

Fixed an issue that may cause an incomplete result set when executing queries involving LEFT- or RIGHT-JOIN operations using the hash-join algorithm with Parallel Query.

Removed the default roles that were unnecessarily created during the upgrade from AMS2 to AMS3.

Fixed an issue that resolves the deadlock when FLUSH STATUS, COM_CHANGE_USER, and SHOW PROCESS LIST are executed concurrently. (Bug#35218030)