Aurora MySQL database engine updates 2024-11-18 (version 3.08.0, compatible with MySQL 8.0.39)

Added three new Amazon CloudWatch metrics to allow users to monitor the InnoDB purge process:

For more information, see Amazon CloudWatch metrics for Amazon Aurora in the Amazon Aurora User Guide.

Added a new CloudWatch metric TransactionAgeMaximum to allow users to identify long-running transactions that might be holding back the purge boundary. For more information, see Amazon CloudWatch metrics for Amazon Aurora in the Amazon Aurora User Guide.

Added three new CloudWatch metrics for out-of-memory (OOM) avoidance:

For more information, see Amazon CloudWatch metrics for Amazon Aurora in the Amazon Aurora User Guide.

Changed three CloudWatch metrics for out-of-memory (OOM) avoidance from running totals to incremental counters:

For more information, see Amazon CloudWatch metrics for Amazon Aurora in the Amazon Aurora User Guide.

Added two global status variables to show the amount of memory used by internal temporary tables: aurora_temptable_ram_allocation and aurora_temptable_max_ram_allocation. These global status variables increase observability and help diagnose issues related to internal temporary table memory usage.

For more information, see Aurora MySQL global status variables in the Amazon Aurora User Guide.

Introduced the new system variable aurora_optimizer_trace_print_before_purge to print optimizer traces to the error log before the server purges the traces from memory. A purge can be triggered based on the thresholds set by the system variables optimizer_trace_offset and optimizer_trace_limit.

Added support for the following DB instance classes:

For more information, see Supported DB engines for DB instance classes in the Amazon Aurora User Guide.

Introduced a new user for binary log (binlog) replication, rdsrepladmin_priv_checks_user. For more information, see Privilege checks user for binary log replication in the Amazon Aurora User Guide.

Fixed an issue where input parameters to Aurora MySQL stored procedures might be handled incorrectly.

CVE-2023-44487

CVE-2024-0853

CVE-2024-20996

CVE-2024-21000

CVE-2024-21013

CVE-2024-21125

CVE-2024-21127

CVE-2024-21129

CVE-2024-21130

CVE-2024-21134

CVE-2024-21135

CVE-2024-21137

CVE-2024-21142

CVE-2024-21157

CVE-2024-21159

CVE-2024-21160

CVE-2024-21162

CVE-2024-21163

CVE-2024-21165

CVE-2024-21166

CVE-2024-21173

CVE-2024-21177

CVE-2024-21179

CVE-2024-21185

CVE-2024-25062

CVE-2024-37371

Fixed a defect that can cause the server to restart unexpectedly after running Data Manipulation Language (DML) commands on a table that already has instantly dropped columns, such that the count of dropped and nondropped columns is greater than 1017.

Fixed an issue that can lead to database log files not being rotated correctly, resulting in increased local storage space usage on a DB instance.

Fixed an issue that could cause the DB instance to stop closing connections under low-memory conditions when aurora_oom_response is enabled, leading to an out-of-memory reboot that could otherwise be avoided.

Fixed an issue that can cause a reader DB instance to restart when freeing memory used for log application.

Fixed an issue in computing internal metrics for full-text search (FTS) indexes that can cause database restarts.

Fixed a community issue that can sometimes produce incorrect query results when a LIMIT clause is used in subqueries along with index condition pushdown.

Fixed an issue that caused a memory access violation leading to releasing a mutex object no longer owned by the thread.

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 can cause an Aurora reader DB instance to restart when write forwarding is enabled.

Fixed an issue where a query containing multiple UNION clauses could allocate a large amount of memory, leading to a DB instance restart.

Fixed an issue that causes the writer DB instance to restart when running a parallel query on a reader DB instance.

Fixed an issue where binlog replication would stall on the replica due to a deadlock encountered on the replica's I/O thread when the FLUSH RELAY LOGS command was run.

Fixed an issue that causes a DB instance restart when handling large GTID sets on a DB cluster with enhanced binlog enabled.

Fixed an issue that can cause a restart on a binlog replica when the in-memory relay log cache is enabled. The in-memory relay log cache is enabled on Aurora MySQL managed binlog replicas when using either single-threaded binary log replication, or multithreaded replication with GTID auto-positioning enabled.

Fixed an issue that can cause a binlog replica instance to restart when applying Data Control Language (DCL) statements during database engine startup.

Fixed an issue that, in rare conditions, can cause a reader DB instance to restart due to a deadlatch when running SELECT queries on tables being updated by its writer DB instance.

Fixed an issue that can cause an Aurora Global Database reader instance to restart with an active write forwarding session.

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 cause a database to restart when scheduled events are canceled while running on DB instances that have enhanced binlog enabled.

Fixed an issue where a client connection can become stuck during zero-downtime patching (ZDP) or zero-downtime restart (ZDR).

Fixed an issue that, in rare cases, cause a database instance restart due to a memory management issue that can occur while handling an open table failure.

Fixed an issue that can cause SQL statements to experience unexpected primary key violation errors or warnings on some rows when performing concurrent INSERT statements on a table that has an AUTO_INCREMENT primary key column and a unique key column, and when an INSERT statement has unique key violations on different rows.

The Performance Schema instrument memory/sql/sp_head::main_mem_root is now a controlled instrument. Consequently, memory allocated for parsing and representation of stored programs now contributes toward per-connection memory limits.

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

Fixed an issue that prevented users from disabling local write forwarding.

Fixed an issue where a SELECT COUNT query can return the wrong result while using write forwarding.

Fixed an issue that can lead to incorrect query results when ZDP incorrectly restores session variables set as hints in queries.

Fixed an issue with automatic truncation of undo tablespaces when they're larger than the threshold innodb_max_undo_log_size in upgrade scenarios.

Fixed an issue where the commit latency and commit throughput aren't measured when innodb_flush_log_at_trx_commit is set to 0.

Fixed an issue that can cause Aurora replica lag to be reported incorrectly .

Fixed an issue that provided an incorrect value for the threads_running variable when using Aurora Global Database.

Fixed an issue where an Aurora MySQL binlog replica with parallel secondary index optimization enabled would experience a restart when applying replication changes on tables with foreign keys.

Fixed a restart issue caused by prolonged resizing of the lock hash table during scaling up or scaling down events.

Fixed an issue that can cause a DB instance restart after disabling binary logging, when enhanced binlog was previously enabled.

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

Fixed an issue where a user might experience an ERROR 1377 (HY000): Fatal error during log purge error while running the mysql.rds_set_external_source stored procedure on a binlog replica, when binary log replication is already configured.

Fixed a defect that prevented the persistence of user role privileges after ZDR.

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

Fixed an issue that can cause a DB instance to restart when SHOW VOLUME STATUS is run.

Fixed a restart issue caused by prolonged buffer pool resizing during scaling up or scaling down events.

Fixed a restart issue caused by logical read ahead (LRA) accessing freed pages due to buffer resizing during a scaling down event.

Fixed an issue that addresses SELECT queries returning incorrect results when the query uses LEFT OUTER JOIN with semijoin transformation that uses materialization as a strategy.

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

Fixed an issue where the ActiveTransactions and BlockedTransactions CloudWatch metrics were reporting lower values than expected.

Fixed an issue where binlog replication breaks when the replica processes a multitable DELETE statement that explicitly deletes from both a parent and a child table.

Fixed an issue that can lead to a DB instance restart when processing a trigger with user-defined functions that return an enum type.

Fixed an issue where the DMLLatency CloudWatch metric would show incorrect values for a binary log replica instance when replicating using binlog_format set to row.

Fixed an issue where slow INSERT, DELETE, and UPDATE queries run by the MySQL Event Scheduler weren't recorded in the slow query log unless preceded by a slow SELECT query.

Fixed an issue that, in rare cases, caused either intermittent unavailability of an Aurora read replica or table definition inconsistencies, sometimes with the error Table does not exist, on the replica. This is due to concurrent read queries on the replica and Data Definition Language (DDL) operations on the writer DB instance.

Fixed an issue that caused the SHOW BINARY LOGS command to take longer to run on a DB 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.

Improved the performance of major version upgrade from Aurora MySQL version 2 to version 3 for DB clusters with large numbers of database objects (such as tables, triggers, and routines).

For larger DB instance classes, the database upgrade process upgrades traditional MySQL object metadata to the new, atomic MySQL 8.0 data dictionary in parallel using multiple threads.

Fixed an issue that causes upgrades or migrations to fail when the target Aurora MySQL DB engine version is 3.04.0 or higher. This occurs when the lower_case_table_names DB cluster parameter is set to 1, and MySQL database collation is incompatible with lowercase table names.

Fixed an issue that caused NULL values to be incorrectly omitted from the result set for certain queries that have both JOIN and UNION operations. (Community Bug Fix #114301)