MariaDB Enterprise Server enhances the community version with advanced features, security, high availability, support, and tools for enterprise use.
MariaDB Enterprise Server - Data-in-Transit Encryption
About
Information about MariaDB Enterprise Server releases
MariaDB Enterprise Server 11.8
Release notes for MariaDB Enterprise Server 11.8 series releases
MariaDB Enterprise Server 11.4
Release notes for MariaDB Enterprise Server 11.4 series releases
MariaDB Enterprise Server 10.6
Release notes for MariaDB Enterprise Server 10.6 series releases
Encryption Plugins
MariaDB's encryption plugins provide transparent data encryption (TDE) for stored data, securing tablespaces and logs to protect sensitive information and meet compliance.
Enterprise Server Release Schedule
The release schedule for MariaDB Enterprise Server
About
MariaDB Enterprise Server enables a predictable development and operations experience through an enterprise lifecycle. This enterprise lifecycle incorporates optimized builds, predictable release behavior, and vendor support.
Minor and maintenance releases of MariaDB Enterprise Server incorporate bug fixes and security fixes. Like all releases of MariaDB Enterprise Server, minor and maintenance releases undergo extensive QA testing before release to customers.
The frequency of minor and maintenance releases is variable. Minor or maintenance releases occur for each release series at least quarterly. A portion of minor and maintenance releases occur on a pre-scheduled basis. Additional releases may be added by MariaDB Corporation between scheduled release dates to manage the flow of change or to address specific defects.
Next Scheduled Releases
Upcoming minor or maintenance releases are scheduled for:
2026-06-15
2026-09-14
2026-12-14
2027-03-15
Prior Releases
See the list of
Obtaining MariaDB Enterprise Server
MariaDB Enterprise Server is available to MariaDB subscription customers via the .
If you are not yet a MariaDB subscription customer, for more information.
MariaDB Enterprise Server binary and source code is delivered over secure protocols from MariaDB Corporation-maintained infrastructure.
MariaDB Enterprise Server 10.6.17-13 is a Stable (GA) maintenance release of 10.6. This release includes a fix for a regression introduced with MariaDB Enterprise Server 10.6.17-12. Users of MariaDB Enterprise Server 10.6.17-12 are encouraged to upgrade to the latest version 10.6.17-13.
MariaDB Enterprise Server 10.6.17-13 was released on 2024-04-24.
Changes in Storage Engines
This release incorporates MariaDB ColumnStore engine version 23.10.1.
Issues Fixed
Can result in hang or crash
An UPDATE to a unique/primary key field can result in a server crash, if the update results in a duplicate key error message. The crash is more likely to happen for a server configuration with a lower setting than the default (299008) for --thread_stack, or when the UPDATE is executed inside of a stored function (MENT-2065)
For MariaDB Enterprise Server 10.6.17-12, if an update is not possible, a workaround is to increase the current setting of --thread_stack by 100k
Platforms
In alignment to the , MariaDB Enterprise Server 10.6.17-13 is provided for:
AlmaLinux 8 (x86_64, ARM64)
AlmaLinux 9 (x86_64, ARM64)
CentOS 7 (x86_64)
Debian 10 (x86_64, ARM64)
Some components of MariaDB Enterprise Server might not support all platforms. For additional information, see
and MariaDB Community Server support data-in-transit encryption, which secures data transmitted over the network. The server and clients encrypt data using the Transport Layer Security (TLS) protocol, a newer version of the Secure Socket Layer (SSL) protocol.
is available on the MariaDB documentation.
Debian 11 (x86_64, ARM64)
Debian 12 (x86_64, ARM64)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
MariaDB Enterprise Server 10.6.23-19 is a Stable (GA) maintenance release of . This release includes a variety of fixes.
Changes in Storage Engines
This release incorporates MariaDB ColumnStore engine version 23.10.6
Notable changes
updated to
Issues Fixed
Can result in data loss
Adding (with ALTER TABLE) a UNIQUE constraint that is USING HASH to a table with foreign keys could've caused the table to become corrupted. ()
Can result in hang or crash
Server crashes in do_mark_index_columns instead of ER_DUP_ENTRY on partitioned table ()
CREATE OR REPLACE with self-referencing CHECK hangs forever, cannot be killed ()
Segfault on INTERSECT ALL with UNION in Oracle mode ()
Can result in unexpected behaviour
ALTER TABLE require ALTER privilege on sequence from DEFAULT value expression ()
In certain cases privileges on sequences were too restrictive, for example, SELECT on a table might've erroneously required INSERT privilege on a sequences ()
AUTO_INCREMENT leads to non-serializable on results ()
Changelog
For the complete list of changes in this release, see the .
Platforms
In alignment to the , MariaDB Enterprise Server 10.6.23-19 is provided for:
AlmaLinux 8 (x86_64, ARM64)
AlmaLinux 9 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Debian 12 (x86_64, ARM64)
Some components of MariaDB Enterprise Server are supported on a subset of platforms. See for details.
This commit fixes a bug where Aria tables are used in (master->slave1->slave2) and a backup is taken on slave2. In this case it is possible that the replication position in the backup, stored in mysql.gtid_slave_pos, will be wrong. This will lead to replication errors if one is trying to use the backup as a new slave. (MDEV-36143)
Replica stops with "Found invalid event in binary log" (MENT-2385)
UNIQUE constraint that was USING HASH and UNIQUE constrant WITHOUT OVERLAPS could be violated under heavy load in READ COMMITTED transaction isolation mode. (MDEV-37199)
UNIQUE KEY USING HASH accepting duplicate records (MENT-2324)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Oracle Linux 8 (x86_64, ARM64)
Oracle Linux 9 (x86_64, ARM64)
Red Hat Enterprise Linux 8 (x86_64, ARM64)
Red Hat Enterprise Linux 9 (x86_64, ARM64, PPC64LE)
Rocky Linux 8 (x86_64, ARM64)
Rocky Linux 9 (x86_64, ARM64)
SUSE Linux Enterprise Server 12 (x86_64)
SUSE Linux Enterprise Server 15 (x86_64, ARM64)
Ubuntu 22.04 (x86_64, ARM64)
Ubuntu 24.04 (x86_64, ARM64)
Red Hat UBI 8 (x86_64, ARM64)
Red Hat UBI 8 is part of the Enterprise Server Docker Image. It does not support MariaDB Enterprise Cluster (Galera) or MariaDB ColumnStore.
MariaDB Enterprise Server 10.6.25-21 is a Stable (GA) maintenance release of MariaDB Enterprise Server 10.6. This release includes a variety of fixes.
Changes in Storage Engines
This release incorporates MariaDB ColumnStore engine version 25.10.4
Notable changes
updated to 26.4.26
Issues Fixed
Can result in data loss
Server crash upon moving InnoDB table with fulltext index between databases ()
When innodb_buffer_pool_size is not large enough to buffer all log records during crash recovery, InnoDB may fail to recover a page from the doublewrite buffer. ()
When an inplace ALTER operation is rolled back, InnoDB drops intermediate tables and their associated FTS internal tables. However, MariaBackup's DDL tracking can incorrectly report this as a backup failure. ()
Can result in hang or crash
Fixed crash in Aria when doing a sub transaction like reading an entry from the proc table. ()
CREATE OR REPLACE ... SELECT where select used functions or sys tables could crash in debug builds, no problems with normal builds ()
innodb_undo_log_truncate=ON leads to out-of-bounds write ()
Can result in unexpected behavior
Bogus [ERROR] InnoDB: Compressed page checksum mismatch could be reported for ROW_FORMAT=COMPRESSED tables. ()
ed25519 client authentication plugin would fail to load for anything but mariadb client utility ()
If the server was killed during an operation that creates a .ibd file, such as TRUNCATE TABLE, it could fail to recover. ()
Related to Install and Upgrade
There was a short gap when mariadb-secure-installation temporary files were potentially openable/readable by an unprivileged user. These files may contain database root password. ()
Related to performance
ASCII strings will be sent to client connections slightly faster. ()
Galera threads time out under heavy load ()
Unexpected results
INSERT ... RETURNING exposes columns for which the user lacks SELECT privilege ()
Presence of Spider disables "external" XA commit ()
Changelog
For the complete list of changes in this release, see the .
Platforms
In alignment to the enterprise lifecycle, MariaDB Enterprise Server 10.6.24-20 is provided for:
AlmaLinux 8 (x86_64, ARM64)
AlmaLinux 9 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Debian 12 (x86_64, ARM64)
Some components of MariaDB Enterprise Server are supported on a subset of platforms. See for details.
Release Notes for MariaDB Enterprise Server 10.4.15-9
This ninth release of 10.4 is a hotfix release to address a CVE which has been rated "critical" using the criteria defined in the .
MariaDB Enterprise Server 10.4.15-9 was released on 2020-10-07.
Fixed Security Vulnerabilities
InnoDB could crash during a workload that is frequently rebuilding or dropping tables by executing statements such as TRUNCATE TABLE, OPTIMIZE TABLE, or DROP TABLE. (MDEV-37755)
Secondary indexes could be corrupted for InnoDB tables that contain indexed virtual columns. (MDEV-38140)
Specially crafted packet could cause the server to crash (MDEV-38242)
The server can hang when data-at-rest encryption is used with multiple encryption threads (MDEV-38271)
When installing mysql-selinux RPM it is no longer possible to re-start Galera nodes (MDEV-38297)
Spider could crash when loop detection variables were artificially manipulated. (MDEV-38368)
malformed spatial data could've caused a crash inside InnoDB (MDEV-38372)
Long blob prefix keys could cause a crash in galera. (MDEV-38374)
we are trying to push aggregate into WHERE (MDEV-38487)
Fixed crash in Aria when doing a sub transaction like reading an entry from the proc table. (MENT-2424)
A replica would crash while replicating UPDATE and DELETE DML statements that target a table which previously had a partition that was converted to a separate table via ALTER TABLE .. CONVERT PARTITION .. TO TABLE. For example, if the command looked like ALTER TABLE t1 CONVERT PARTITION p1 TO TABLE t_new; the replica would crash when trying to update/deleterows in table t1 after running the command. (MENT-2466)
With --encrypt-binlog=ON if a node fails to apply a writeset it will crash the whole cluster due to a bug in Galera library encryption handling. Fixed in Galera library (MENT-2474)
SET GLOBAL innodb_buffer_pool_size was prone to hang or crash the server, especially when innodb_adaptive_hash_index was enabled. (MENT-2545)
If the server was killed during an operation that creates a .ibd file, such as TRUNCATE TABLE, it could fail to recover. (MDEV-38026)
Auth Switch with Long Password Corrupts Database Name (MDEV-38431)
Oracle Linux 8 (x86_64, ARM64)
Oracle Linux 9 (x86_64, ARM64)
Red Hat Enterprise Linux 8 (x86_64, ARM64)
Red Hat Enterprise Linux 9 (x86_64, ARM64, PPC64LE)
Rocky Linux 8 (x86_64, ARM64)
Rocky Linux 9 (x86_64, ARM64)
SUSE Linux Enterprise Server 12 (x86_64)
SUSE Linux Enterprise Server 15 (x86_64, ARM64)
Ubuntu 22.04 (x86_64, ARM64)
Ubuntu 24.04 (x86_64, ARM64)
Microsoft Windows (x86_64) (Without MariaDB Enterprise Cluster (Galera) support)
Red Hat UBI 8 (x86_64, ARM64)
Red Hat UBI 8 is part of the Enterprise Server Docker Image. It does not support MariaDB Enterprise Cluster (Galera) or MariaDB ColumnStore.
CentOS 6, Debian 8, and Red Hat Enterprise Linux 6 are no longer supported as per the MariaDB Engineering Policy. Older releases are available from the MariaDB Downloads page. Instructions for installation are included as a README file within the download.
MariaDB Enterprise Server 10.6.12-8 is a Stable (GA) maintenance release of 10.6. This release includes a variety of fixes. Users of MariaDB Enterprise Server 10.6.12-7 are encouraged to upgrade. are required if upgrading from Enterprise Server 10.6.12-7 to 10.6.12-8.
The next for MariaDB Enterprise Server is 2023-06-12.
MariaDB Enterprise Server 10.6.12-8 was released on 2023-05-24.
Fixed Security Vulnerabilities
#1:
MariaDB CVEs are assigned a word rating instead of a CVSS base score. See the for details.
Upgrade Procedure
MariaDB Enterprise Server 10.6.12-8 has a special upgrade procedure that only applies to users who previously deployed MariaDB Enterprise Server 10.6.12-7.
The following procedure is required to truncate the undo log tablespaces:
Upgrade from ES 10.6.12-7 to 10.6.12-8 using the normal upgrade procedure.
After the server has been started with ES 10.6.12-8, set innodb_fast_shutdown=0:
Shutdown the server:
Change innodb_undo_tablespaces to a different value than its current value:
Start the server:
Notable Changes
The optimizer can now use column histogram data when estimating output cardinality of hashed (BNL-H) joins. ()
Starting with this release, the has a new flag called hash_join_cardinality that can enable the feature.
When enabled, the optimizer uses a formula that provides conservative estimates.
Issues Fixed
Can result in a hang or crash
When a prepared statement is used to query a view that contains a UNION, the server can crash. ()
When is set and is greater than 0, replica servers can hang if data is inserted into tables with a sequence. ()
Can result in unexpected behavior
For certain data distributions, the optimizer histogram code can produce wrong selectivity, which can lead to performance degradation. ()
The optimizer does not take into account the selectivity of the equality conditions for Block Nested Hash (BNL-H) joins.
In previous releases, this issue can cause mis-estimates and bad query plans when running with set to 3 or higher (the default is 2).
InnoDB does not free undo logs when they are no longer needed. ()
In the previous release, when is set to 0, the undo logs would be stored in the InnoDB system tablespace (ibdata1 by default), so this issue could cause the InnoDB system tablespace to grow indefinitely.
Starting with this release, InnoDB frees undo logs when they are no longer needed.
Changes in Storage Engines
This release incorporates MariaDB ColumnStore engine version 23.02.3.
Interface Changes
hash_join_cardinality=off has been incrementally added to the default value of the system variable.
Platforms
In alignment to the , MariaDB Enterprise Server 10.6.12-8 is provided for:
CentOS 7 (x86_64)
Debian 10 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Red Hat Enterprise Linux 7 (x86_64)
Windows packages are not currently available for this release.
Some components of MariaDB Enterprise Server might not support all platforms. For additional information, see
Release Notes for MariaDB Enterprise Server 10.4.34-24
MariaDB Enterprise Server 10.4.34-24 is a maintenance release of . This release includes a variety of fixes.
MariaDB Enterprise Server 10.4.34-24 was released on 2024-06-11.
Notable Changes
Galera has been updated to 26.4.18
The GCS protocol version has been changed, preventing a downgrade of individual nodes of a MariaDB Enterprise Cluster
Issues Fixed
Can result in data loss
With --gtid-ignore-duplicate set, a transaction can be double-applied from another replication source if applying the transaction from the initial source required retrying in parallel execution. ()
Backups of server with innodb_encrypt_tables=1 can become corrupted in mariadb-backup --prepare ()
Can result in a hang or a crash
Using current MariaDB Enterprise Backup against an older server can result in a crash, as the system variable @@aria_log_dir_path does not exist ()
For encrypted table spaces, an ALTER operation can hang when an encryption thread works on the same tablespace ()
EXPLAIN statement that uses a subquery which has a query plan that A) will examine less than @@expensive_subquery_limit rows and B) will use the join buffer, could cause a crash. (
Can result in unexpected behavior
Spider/ODBC passed double quotes for names, in ANSI style ()
Default charset doesn't work with the PHP MySQLi extension ()
Spider returns parsing failure on valid left join select by translating the on expression to () ()
Related to performance
Table is getting rebuilt with ALTER TABLE ADD COLUMN, although it should be an instant operation, not requiring a rebuild ()
Changelog
For the complete list of changes in this release, see the .
Platforms
In alignment with the , MariaDB Enterprise Server 10.4.34-24 is provided for:
CentOS 7 (x86_64)
Debian 10 (x86_64)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Red Hat Enterprise Linux 7 (x86_64)
Some of the components of MariaDB Enterprise Server are supported on a subset of platforms. See for details.
To enable this functionality in a given session, set the flag using the :
In previous releases, the replica server would treat full inserts into the sequence as DDL statements, which would cause it to acquire an exclusive lock on the sequence table. If another parallel replication thread is waiting on the lock at the same time, the two threads can deadlock.
Starting with this release, replica servers manage their locks better to avoid deadlocks in this scenario.
When InnoDB splits or merges a B-tree page, the server can hang due to a race condition. (MDEV-29835)
In previous releases, InnoDB chooses an undo log tablespace for truncation and marks all the related rollback segments with the state "truncation in progress". When the truncation is completed, InnoDB fails to re-enable the rollback segments. Consequently, if there are further write transactions before the server is shut down, the transactions must wait for usable rollback segments in a busy-loop. Since InnoDB does not re-enable the rollback segments, the transactions wait indefinitely.
Starting with this release, InnoDB properly manages the state of rollback segments when InnoDB undo log truncation is enabled.
When the lateral derived optimization is used to execute a query, the derived table is re-filled on every incoming row combination, even if the parameter values have not changed. (MDEV-26301)
When is optimistic and is greater than 0, can fail with an out-of-order binlog error if the sequence uses InnoDB. (MDEV-31077)
In previous releases, the following error can be raised:
For servers that previously ran ES 10.6.12-7, a special procedure is required to truncate the undo log tablespaces after the upgrade is complete.
InnoDB history list length increases faster in MariaDB Enterprise Server 10.6 compared to MariaDB Enterprise Server 10.5 for the same load. (MDEV-29401)
Galera-replicated events can, in some cases, contain the wrong time when versioning is used (MDEV-18590)
)
MariaDB Enterprise Backup fails with the following error message if the prepare step of the backup encounters a data directory, which happens to store wsrep xid position in TRX SYS page: (MDEV-33540)
Original IP not shown in network-related error messages when proxy_protocol is in use (MDEV-33506)
Incorrect DEFAULT expression evaluated in UPDATE (MDEV-33790)
Last_Error: Error 'An attempt was made to binlog GTID 0-1-100 which would create an out-of-order sequence number with existing GTID 0-1-100 and gtid stric mode is enabled' on query. Default database: 'test'. Query: 'alter sequence s1 restart with 1' will be shown.
InnoDB: Crash recovery is broken due to insufficient innodb_log_file_size
Release Notes for MariaDB Enterprise Server 11.4.8-5
MariaDB Enterprise Server 11.4.8-5 is a Stable (GA) maintenance release of MariaDB Enterprise Server 11.4, released on 2025-09-08
The most recent release of MariaDB Enterprise Server 11.4 is:
MariaDB Enterprise Server 11.4.8-5 is a Stable (GA) maintenance release of . This release includes a variety of fixes.
Changes in Storage Engines
This release incorporates MariaDB ColumnStore engine version 23.10.6
Backports
Segmented key cache for Aria ()
A new variable (default 1) to define the number of segments has been added. The default disables the new feature.
Backport fixes for Vectors to MariaDB Enterprise 11.4 ()
Notable changes
Repositories for RedHat 10 x86_64 and aarch64 have been added
Support added for Red Hat Enterprise Linux 10, AlmaLinux 10, Rocky Linux 10 and Oracle Linux 10
MariaDB systemd multi-instance service was changed to not attempt changes to the permissions on its pam helper server. This prevents unconstructive behavior and errors in the systemd journal when starting the mariadb@.service ()
Issues Fixed
Can result in data loss
Adding (with ALTER TABLE) a UNIQUE constraint that is USING HASH to a table with foreign keys could've caused the table to become corrupted. ()
Can result in hang or crash
Server crashes in do_mark_index_columns instead of ER_DUP_ENTRY on partitioned table ()
CREATE OR REPLACE with self-referencing CHECK hangs forever, cannot be killed ()
Segfault on INTERSECT ALL with UNION in Oracle mode ()
Can result in unexpected behaviour
After a successful shrinking of innodb_buffer_pool_size, there will be no message about it in the server error log. After a failed shrinking of innodb_buffer_pool_size, the adaptive hash index will not be re-enabled if innodb_adaptive_hash_index was ON when SET GLOBAL innodb_buffer_pool_size=... started to execute. ()
ALTER TABLE require ALTER privilege on sequence from DEFAULT value expression ()
Deadlock does not rollback transaction fully ()
Unexpected results
Remove the error codes added by the MDEV-36032 patch ()
Seconds_Behind_Master Spike at Log Rotation on Parallel Replica ()
Assertion 'vers_conditions.delete_history' failed upon PREPARE ()
Related to performance
Performance regression with encrypted InnoDB log ()
Changelog
For the complete list of changes in this release, see the .
Platforms
In alignment to the enterprise lifecycle, MariaDB Enterprise Server 11.4.8-5 is provided for:
AlmaLinux 8 (x86_64, ARM64)
AlmaLinux 9 (x86_64, ARM64)
AlmaLinux 10 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Some components of MariaDB Enterprise Server are supported on a subset of platforms. See for details.
MariaDB Enterprise Server 10.6.12-7 is a Stable (GA) maintenance release of 10.6. This release includes a variety of fixes.
MariaDB Enterprise Server 10.6.12-7 was released on 2023-03-13.
Users of MariaDB Enterprise Server 10.6.12-7 are encouraged to upgrade to 10.6.12-8. Please see the that must be used.
Backported Features
MariaDB Enterprise Server enables a predictable development and operations experience through an . These new features have been backported after reaching maturity in MariaDB Community Server.
In previous releases, the number of undo logs was configurable before InnoDB was initialized. With this release, the number of undo logs can also be configured after install.
The number of undo logs is configured by the InnoDB system variable --innodb-undo-tablespaces.
Splitting undo logs over multiple tablespaces can reduce the size of a single tablespace, especially the InnoDB system tablespace.
Notable Changes
Output for the function has been optimized to reduce the number of lines needed. ()
For compatibility, has been added as an alias to JSON_DETAILED(). ()
This release incorporates Galera library version 26.4.14.
Starting with this release, the mariadb-backup execution can be aborted if DDL statements happen during backup execution. This abortable backup execution is the optional feature which can be enabled/disabled by wsrep_mode: BF_ABORT_mariadb-backup
Issues Fixed
Can result in data loss
When executing a DELETE or UPDATE with a subselect, the server can crash. ()
When performing an incremental backup, MariaDB Enterprise Backup does not reflect databases that have been dropped and created. ()
During recovery, MariaDB Server can crash with an error when a DDL operation was executed before the server was stopped: InnoDB: Trying to write ... bytes at ... outside the bounds of the file ...
Can result in a hang or crash
When DELETE HISTORY is executed for a system versioned table with full text index, the server can crash. ()
When an ALTER TABLE statement causes InnoDB to rebuild a table with a spatial index, the server can crash. ()
When using SPIDER with SPIDER_DIRECT_SQL
Can result in unexpected behavior
log_query_not_using_indexes=OFF is ignored when log_slow_filter is an empty string. ()
MariaDB Enterprise Backup returns with an error when the option --galera-info is used for creating a backup of a MariaDB Server instance, which is not a MariaDB Enterprise Cluster node. Failed to get master wsrep state from SHOW STATUS ()
()
When running mariadb-binlog using the option --verbose, cannot read row events with compressed columns: Error: Don't know how to handle column type:... ()
Incorrect results are returned with outer join, merged derived table, and view. ()
Some DDL, such as ANALYZE
Changes in Storage Engines
This release originally incorporated MariaDB ColumnStore storage engine version 23.02.1.
This release later incorporated MariaDB ColumnStore storage engine version 23.02.2.
This release now incorporates MariaDB ColumnStore storage engine version 23.02.3.
Interface Changes
Error code 5016 added.
system variable default value changed from ON to OFF
function added.
Platforms
In alignment to the , MariaDB Enterprise Server 10.6.12-7 is provided for:
CentOS 7 (x86_64)
Debian 10 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Some components of MariaDB Enterprise Server might not support all platforms. For additional information, see
MariaDB Enterprise Server 10.6.22-18 is a Stable (GA) maintenance release of 10.6. This release includes a variety of fixes.
MariaDB Enterprise Server 10.6.22-18 was released on 11 Jun 2025.
Fixed Security Vulnerabilities
Changes in Storage Engines
This release incorporates MariaDB ColumnStore engine version .
Notable changes
New user variable, analyze_max_length, default value 4G. Any field that is bigger than this value in bytes will be ignored by ANALYZE TABLE PERSISTENT to not collect statistics for long char/varchars unless it is specified in FOR COLUMNS(). (MENT-2301)
SBOM now includes 'license' and 'copyright' information, improving security compliance and dependency tracking. (MENT-2311)
MariaDB effectively running as root CAP_DAC_OVERRIDE ()
Issues Fixed
Can result in data loss
Assertion InnoDB searching row in wrong partition for multiple system versioned DELETE with same timestamp and same multistatement transaction ()
Can result in hang or crash
Incorrect error handling on DDL with FULLTEXT INDEX ()
ALTER TABLE…SEQUENCE does not work correctly with InnoDB ()
Race conditions between ALTER TABLE or OPTIMIZE TABLE and the purge of transaction history were fixed. ()
Can result in unexpected behaviour
CREATE INDEX fails to heal a FOREIGN KEY constraint ()
Doublewrite recovery of innodb_checksum_algorithm=full_crc32 page_compressed pages does not work ()
Segfault on concurrent ALTER and SELECT for partitioned table ()
Unexpected results
The untested ha_spider::index_first_internal constructs broken queries ()
Wrong results from tables with a single record and an aggregate ()
Unexpected error 1264 'Out of Range Value for Column' when inserting into ... select ... from a spider table ()
Release Notes for MariaDB Enterprise Server 10.5.27-21
MariaDB Enterprise Server 10.5.27-21 is a Stable (GA) maintenance release of MariaDB Enterprise Server 10.5, released on 2024-12-10
MariaDB Enterprise Server 10.5.27-21 is a maintenance release of MariaDB Enterprise Server 10.5. This release includes a variety of fixes.
MariaDB Enterprise Server 10.5.27-21 was released on 2024-12-10.
Notable Changes
A new parameter --quick-max-column-width is now available in the mariadb client to limit the field width when used with the --quick mode ()
Changes in Storage Engines
This release incorporates MariaDB ColumnStore engine version 5.6.8
Issues Fixed
Can result in data loss
Using group_concat with a compressed or GIS column can lead to a server crash and potential data corruption for users performing group_concat operations on tables containing these data types ()
Can result in hang or crash
A query could cause a crash if it has a HAVING clause with a construct tblX.column=column_or_constant and the optimizer was able to infer that table tblX is a constant table. Note that HAVING clause may be from the original query or may come from ) ()
Calling a UDF for the engine SPIDER results in a crash if the SPIDER storage engine plugin could not be loaded ()
Under Windows Subsystem for Linux, InnoDB crashes on ALTER TABLE or OPTIMIZE TABLE ()
In some situations, where a MariaDB Enterprise cluster (Galera) node crashes, some threads are still working. This is visible via the error log, which is still getting errors written by the node. As the node stops interacting correctly with other cluster nodes and is blocking them from taking over the primary state, the whole cluster hangs ()
Can result in unexpected behavior
When a user runs mariadb-binlog with --stop-position, they would expect the output to contain events up to that event. If the output did not contain events up to that event, this may result in various unexpected behaviors, e.g., an incomplete database state if they piped the output to the mariadb client and expected certain transactions to have executed in the database, yet never were run ()
Changing a data type of a field used in a foreign key constraint fails with Error "Cannot change column '...': used in a foreign key constraint '...'" ()
Related to performance
Unnecessary copying of log records by MariaDB Enterprise Backup when further transaction commits are blocked by BACKUP STAGE BLOCK_COMMIT (MENT-2133)
It also would cause further effort of rolling back incomplete transactions after the backup is restored
Aria internal temporary tables unnecessarily write all changed blocks to disk when the table is closed at end of query (MENT-2182)
Changelog
For the complete list of changes in this release, see the .
Platforms
In alignment to the , MariaDB Enterprise Server 10.5.27-21 is provided for:
AlmaLinux 8 (x86_64, ARM64)
AlmaLinux 9 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Some components of MariaDB Enterprise Server might not support all platforms. For additional information, see "".
Release Notes for MariaDB Enterprise Server 11.4.10-7
MariaDB Enterprise Server 11.4.10-7 is a Stable (GA) maintenance release of MariaDB Enterprise Server 11.4, released on 2026-03-17
Release date: 17 Mar 2026
MariaDB Enterprise Server 11.4.10-7 is a Stable (GA) maintenance release of . This release includes a variety of fixes.
Changes in Storage Engines
Release Notes for MariaDB Enterprise Server 10.6.18-14
MariaDB Enterprise Server 11.6.18-14 is a Stable (GA) maintenance release of MariaDB Enterprise Server 10.6, released on 2024-06-11
The most recent release of MariaDB Enterprise Server 10.6 is:
Release Notes for MariaDB Enterprise Server 10.6.20-16
MariaDB Enterprise Server 11.6.20-16 is a Stable (GA) maintenance release of MariaDB Enterprise Server 10.6, released on 2024-12-10
The most recent release of MariaDB Enterprise Server 10.6 is:
Field pointer may be uninitialized in fill_record (MDEV-36181)
Incorrect undo logging for indexes on virtual columns whose index ID does not fit in 32 bits (MDEV-36613)
With wsrep_ignore_apply_errors = 0, the node crashes due to assertion thd->is_error() failed in Sql_cmd_dml::prepare(), shown in the logs (MDEV-35946)
In some cases, if there are MDL locks (for example, when LOCK TABLE is executed), a node could get stuck in the system thread due to incorrect handling of metadata locks (MDL) in server code when a transaction was BF aborted. (MDEV-35941)
Regression after the fix for MDEV-31413 - sometimes the server crashes with an assertion in wsrep::transaction::before_rollback(), for example when using OPTIMIZE TABLE on an ARIA table with wsrep_osu_method=RSU. (MDEV-32631)
SST failure occurs when gtid_strict_mode is enabled under high load, such as OLTP load is active on the primary node. A typical symptom of this error is the presence of the diagnostic "[ERROR] mariadbd: Error writing file 'binlog'", in the debug version it is also possible to crash on assertion in the wsrep::transaction::before_rollback() function with the message "Assertion `state() == s_executing || state() == s_preparing || state() == s_prepared || state() == s_must_abort || state() == s_aborting || state() == s_cert_failed || state() == s_must_replay' failed". (MDEV-34891)
In Galera, creating sequence with a small cache leads to signal 6 error: [ERROR] WSREP: FSM: no such a transition REPLICATING -> COMMITTED. (MDEV-33850)
Under high load wsrep internal thread may terminate due to memory pressure conditions, but this is not a crash, however in debug version user may encounter assertion in wsrep_to_isolation_begin() function with following message: "int wsrep_to_isolation_begin(THD*, const char*, const char*, const TABLE_LIST*, const Alter_info*, const key_array*, const HA_CREATE_INFO*): Assertion `(0)' failed." (MDEV-36116)
Assertion `commit_trx' failed in innobase_commit() (ha_innodb.cc). An INSERT with sql_log_bin=0 is still replicated in Galera (per MDEV-7205), despite binary logging being disabled. This results in a partial binlog bypass, requiring a two-phase commit (2PC). During 2PC, the INSERT is first prepared (entering the PREPARED state in InnoDB), and on commit, the new assertion from MDEV-24035 fails, causing a crash with "Assertion 'commit_trx' failed" in logs. (MDEV-35658)
When a sequence is used and inserts run in parallel on multiple Galera nodes, a transaction may be aborted after passing certification. If it then attempts to roll back, the binlog statement cache—which includes reserved sequence values—may be written prematurely. This causes a crash with the diagnostic "WSREP: FSM: no such a transition REPLICATING -> COMMITTED" in the logs, as the transaction is supposed to replay and only write to the binlog during the final commit. (MDEV-33589)
corruption when query cache cannot allocate block (MDEV-34075)
Stack looping and SIGSEGV in Item_args::walk_args on UPDATE (MDEV-31647)
Server crash in find_field_in_tables, Assertion `name' failed in find_field_in_table_ref (MDEV-25012)
Long server_audit_file_path causes buffer overflow (MDEV-36245)
Server crash when inserting from derived table containing insert target table (MDEV-32086)
After a corrupted table on one node triggers the cluster to vote to evict a node that failed a transaction, the current master can’t commit any more and hangs. (MDEV-34998, MENT-2081)
MDEV-32157 intended to fix spider wrapper so that it is case insensitive, among other things. However that fix was incomplete, as the udf spider_direct_sql may still require case sensitivity. MDEV-35807 fixes this. (MDEV-35807)
Creating partitioned tables is disallowed when wsrep_osu_method=TOI and wsrep_strict_ddl=ON, preventing alteration or deletion of partitioned tables. (MDEV-27861)
Attempting to create a CONNECT engine table results in "non-InnoDB sequences in Galera cluster" error message in logs due to an incorrect engine check. (MDEV-35748)
MariaDB Backup returns with an error like "Error on file ./test/t1#P#p513.MYD open during `test`.`t1` table copy for partitioned MyISAM tables when running out of file handles (MENT-2089)
User without any privileges to a sequence can read from it and modify it via column default (MDEV-36413)
User has unauthorized access to a sequence through a view with security invoker (MDEV-36380)
group by handler missing constant fields when selecting from a view (MDEV-36307)
Tests calling the udf spider_copy_tables fail with --view-protocol (MDEV-36335)
Red Hat Enterprise Linux 8 (x86_64, ARM64)
Red Hat Enterprise Linux 9 (x86_64, ARM64)
Rocky Linux 8 (x86_64, ARM64)
Rocky Linux 9 (x86_64, ARM64)
SUSE Linux Enterprise Server 12 (x86_64)
SUSE Linux Enterprise Server 15 (x86_64, ARM64)
Ubuntu 20.04 (x86_64, ARM64)
Red Hat UBI 8 (x86_64, ARM64)
Red Hat UBI 8 is part of the Enterprise Server Docker Image. It does not support MariaDB Enterprise Cluster (Galera) or MariaDB ColumnStore.
This commit fixes a bug where Aria tables are used in (master->slave1->slave2) and a backup is taken on slave2. In this case it is possible that the replication position in the backup, stored in mysql.gtid_slave_pos, will be wrong. This will lead to replication errors if one is trying to use the backup as a new slave. (MDEV-36143)
Maria-backup would crash during the 'maria_recovery' part. This could happen if server was doing repair or creating indexes while the backup was running. (MDEV-36860)
MariaDB server crash when a query includes a derived table containing unnamed column. (MDEV-24588)
Incorrect handling of UTF-8 characters (and other character sets requiring more than a single byte representation) in the minimum/maximum positions of a table during the execution of ANALYZE TABLE tbl PERSISTENT FOR ALL resulted in an endless loop consuming more memory and prevented the server from terminating. (MDEV-36765)
ASAN errors in Binary_string::alloced_length / reset_stmt_params (MDEV-32694)
semi sync makes the master unresponsive when a replica is stopped (MDEV-36934)
json_extract crashes in Item_func_json_extract::read_json (MDEV-21530)
MariaDB 10.11.9 Signal 11 crash on second Stored Procedure call (MDEV-36814)
When innodb_buffer_pool_size is being shrunk while there are no data pages cached, InnoDB could hang or crash. (MDEV-37263)
A replica would crash while replicating UPDATE and DELETE DML statements that target a table which previously had a partition that was converted to a separate table via ALTER TABLE .. CONVERT PARTITION .. TO TABLE. For example, if the command looked like ALTER TABLE t1 CONVERT PARTITION p1 TO TABLE t_new; the replica would crash when trying to update/deleterows in table t1 after running the command. (MDEV-36906)
Crash in add_keyuses_for_splitting() when joining with a derived table (MDEV-30711)
MariaDB Enterprise Backup failed after upgrade (MENT-2401)
Possible crash on server start when CREATE SERVER has been used before (MENT-2388)
Replica stops with "Found invalid event in binary log" (MENT-2385)
parallel slave ALTER-SEQUNCE attempted to binlog out-of-order (MDEV-35570)
In certain cases privileges on sequences were too restrictive, for example, SELECT on a table might've erroneously required INSERT privilege on a sequences (MDEV-36870)
AUTO_INCREMENT leads to non-serializable on results (MDEV-36330)
UNIQUE constraint that was USING HASH and UNIQUE constrant WITHOUT OVERLAPS could be violated under heavy load in READ COMMITTED transaction isolation mode. (MDEV-37199)
Fresh MariaDB 11.4 installation gives errors when configuring utf8 (MDEV-36815)
Split Materialized code: last_refills is never set in 11.0+ (MDEV-36323)
UNIQUE KEY USING HASH accepting duplicate records (MENT-2324)
Crash recovery after bulk load data reports corruption (MDEV-37192)
DML committed within XA transaction block after deadlock error and implicit rollback (MDEV-37141)
Don't generate index_merge plans, if a column is present in both Secondary Key and Primary Key indexes and either of the indexes include it with DESC order (as the scan itself is not a ROR scan). (MDEV-36410)
Incorrect handling of null values on join conditions. (MDEV-37057)
DROP DEFAULT makes SHOW CREATE non-idempotent (MDEV-29001)
Possible wrong results when split materialized optimization is used with nullable join conditions. (MDEV-37230)
Debian 12 (x86_64, ARM64)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Oracle Linux 8 (x86_64, ARM64)
Oracle Linux 9 (x86_64, ARM64)
Oracle Linux 10 (x86_64, ARM64)
Red Hat Enterprise Linux 8 (x86_64, ARM64)
Red Hat Enterprise Linux 9 (x86_64, ARM64, PPC64LE)
Red Hat Enterprise Linux 10 (x86_64, ARM64)
Rocky Linux 8 (x86_64, ARM64)
Rocky Linux 9 (x86_64, ARM64)
Rocky Linux 10 (x86_64, ARM64)
SUSE Linux Enterprise Server 15 (x86_64, ARM64)
Ubuntu 22.04 (x86_64, ARM64)
Ubuntu 24.04 (x86_64, ARM64)
Microsoft Windows (x86_64) (Without MariaDB Enterprise Cluster (Galera) support)
Red Hat UBI 8 (x86_64, ARM64)
Red Hat UBI 8 is part of the Enterprise Server Docker Image. It does not support MariaDB Enterprise Cluster (Galera) or MariaDB ColumnStore.
In previous releases, the number of undo logs had to be configured before InnoDB was initialized, so changing the number of undo logs would require setting up a new server instance. With this backport from MariaDB Community Server 10.11, the number of undo logs can be changed so it will take effect with the next server start. (MENT-1650)
The names used by information_schema.INNODB_TABLESPACES_ENCRYPTION.NAME for undo tablespaces have been changed. (MDEV-30119)
Starting with this release, names for the undo tablespaces use the format undo001.
In previous releases, names for the undo tablespaces used NULL.
For MariaDB Enterprise Cluster SST with MariaDB Enterprise Backup backup execution (namely the block DDL phase) has been changed. In previous releases, the node was paused from the cluster, generating the following log file entry: (MDEV-26391)
(
)
innodb_undo_log_truncate=ON is not crash safe. This also can have effects for MariaDB Enterprise Backup for backup and restore. (MDEV-29999, MDEV-30179, MDEV-30438)
When innodb_undo_log_truncate=ON a backup for MariaDB Enterprise Server can return the error message:
and
spider_udf_ds_use_real_table=1
, the server can crash. (
)
When executing REPLACE INTO ... PARTITION(...), the server can crash. (MDEV-29636)
Galera SST doesn't properly handle undo* files from InnoDB. As a result, SST may terminate incorrectly when --innodb-undo-tablespaces is set to 3 or more. (MDEV-30157)
With a query containing nested WINDOW clauses, the server can crash. (MDEV-30052)
When executing a SELECT from complex a view with WHERE clause and with the setting derived_merge=on, the server can crash. (MDEV-30081)
Infinite sequence of recursive calls when processing embedded CTE, when such a reference has the same table name as the name of the CTE. (MDEV-30248)
With a cluster wide conflict, MariaDB Enterprise Cluster can hang if a binlog purging is running (such as in the case that, binlog expiration is configured). (MDEV-29512)
mariadbd hangs when running with --wsrep-recover and --plugin-load-add=ha_spider.so options. (MDEV-30370)
After a second execution via a stored routine or a prepared statement of a LEAD ... OVER query, the server can crash. (MDEV-28206)
When innodb_undo_log_truncate=ON is set, the server can hang. (MDEV-30180)
After recovery, the server can crash with an error if innodb_buffer_pool_size is not large enough to recover in a single batch: InnoDB: Tried to read ... bytes at offset (MDEV-30132)
Race condition in foreign key locking during truncate. (MDEV-29504)
MariaDB Enterprise Backup can hang when using the parameters --backup --incremental --throttle=... (MDEV-29896)
While executing an ALTER TABLE for a compressed InnoDB table, the server can crash. (MDEV-28797)
InnoDB fails to remove a newly created table or index from data dictionary and table cache and then crashes, if the ALTER to add a FULLTEXT index fails in the commit phase. (MDEV-30393)
When the query cache is enabled and a DROP DATABASE is called, the DROP DATABASE can hang and 100 percent CPU load can be seen. (MDEV-29760)
The execution of FLUSH TABLES...FOR EXPORT can result in a server crash. (MDEV-30227)
When querying a Spider table from a stored procedure inside a WHILE clause, the server can crash. The stored procedure needs to use two WHILE clauses which both query the Spider table. (MDEV-30191)
Deadlock on Replica during BACKUP STAGE BLOCK_COMMIT on XA transactions. (MDEV-30423)
MariaDB Server can crash during recovery if an ALTER TABLE .. ROW_FORMAT=COMPRESSED has been executed before. (MDEV-30404)
When querying a table with virtual generated columns using full text search, these columns are not generated and are always NULL in the result set. (MDEV-29169)
Identifiers are not quoted for the output of SHOW GRANTS. Using the result to execute the grant statement results in a syntax error, if reserved keywords are used. (MDEV-30056)
Spider table with CHARSET utf32/utf16/ucs2 tries to set client CHARSET to unsupported value. (MDEV-29562)
Incorrect results are returned when using STDDEV_SAMP() with a view. (MDEV-19071)
A spurious error can be generated: ERROR 1292 (22007) at the line 15: Truncated incorrect DECIMAL value:
, can be completed out of order on parallel replicas. (
)
seconds_behind_master is incorrect for delayed parallel replicas. (MDEV-29639)
Unlimited SELECT .. ORDER BY .. LIMIT can be slow as it is always using temporary. (MDEV-29129)
Calling DETERMINISTIC package functions from other SQL/PL stored routines can be slow. (MDEV-29370)
An upgrade to MariaDB Server 10.5 and MariaDB 10.6 incorrectly reports that a recovery is needed. (MDEV-24412)
Warning message when encryption is not available on recovery: InnoDB: We do not continue the crash recovery, because the table may become corrupt (MDEV-30068)
With MariaDB Enterprise Backup on Windows, an incremental prepare fails when innodb_undo_tablespaces > 0 is set. (MDEV-30144)
Changing the value of wsrep_gtid_domain_id for MariaDB Enterprise Cluster with full cluster restart can fail on nodes, which do not sync from the node where wsrep_gtid_domain_id was changed. (MDEV-29171)
ASAN errors in dict_sys_t::load_table / get_foreign_key_info after failing to load a table (MDEV-33167)
ALTER TABLE…DROP COLUMN after a failed ALTER TABLE…DROP COLUMN could lead to a server crash (MDEV-36236)
Field pointer may be uninitialized in fill_record (MDEV-36181)
A primary server could crash when a semi-sync connection is stopped, if the primary previously disabled semi-sync replication while the connection was already up (and `rpl_semi_sync_master_wait_no_slave=0`). (MDEV-36359)
Incorrect undo logging for indexes on virtual columns whose index ID does not fit in 32 bits (MDEV-36613)
With wsrep_ignore_apply_errors = 0, the node crashes due to assertion thd->is_error() failed in Sql_cmd_dml::prepare(), shown in the logs (MDEV-35946)
In some cases, if there are MDL locks (for example, when LOCK TABLE is executed), a node could get stuck in the system thread due to incorrect handling of metadata locks (MDL) in server code when a transaction was BF aborted. (MDEV-35941)
Regression after the fix for MDEV-31413 - sometimes the server crashes with an assertion in wsrep::transaction::before_rollback(), for example when using OPTIMIZE TABLE on an ARIA table with wsrep_osu_method=RSU. (MDEV-32631)
SST failure occurs when gtid_strict_mode is enabled under high load, such as OLTP load is active on the primary node. A typical symptom of this error is the presence of the diagnostic "[ERROR] mariadbd: Error writing file 'binlog'", in the debug version it is also possible to crash on assertion in the wsrep::transaction::before_rollback() function with the message "Assertion `state() == s_executing || state() == s_preparing || state() == s_prepared || state() == s_must_abort || state() == s_aborting || state() == s_cert_failed || state() == s_must_replay' failed". (MDEV-34891)
In Galera, creating sequence with a small cache leads to signal 6 error: [ERROR] WSREP: FSM: no such a transition REPLICATING -> COMMITTED. (MDEV-33850)
Under high load wsrep internal thread may terminate due to memory pressure conditions, but this is not a crash, however in debug version user may encounter assertion in wsrep_to_isolation_begin() function with following message: "int wsrep_to_isolation_begin(THD*, const char*, const char*, const TABLE_LIST*, const Alter_info*, const key_array*, const HA_CREATE_INFO*): Assertion `(0)' failed." (MDEV-36116)
Assertion `commit_trx' failed in innobase_commit() (ha_innodb.cc). An INSERT with sql_log_bin=0 is still replicated in Galera (per MDEV-7205), despite binary logging being disabled. This results in a partial binlog bypass, requiring a two-phase commit (2PC). During 2PC, the INSERT is first prepared (entering the PREPARED state in InnoDB), and on commit, the new assertion from MDEV-24035 fails, causing a crash with "Assertion 'commit_trx' failed" in logs. (MDEV-35658)
When a sequence is used and inserts run in parallel on multiple Galera nodes, a transaction may be aborted after passing certification. If it then attempts to roll back, the binlog statement cache—which includes reserved sequence values—may be written prematurely. This causes a crash with the diagnostic "WSREP: FSM: no such a transition REPLICATING -> COMMITTED" in the logs, as the transaction is supposed to replay and only write to the binlog during the final commit. (MDEV-33589)
A Galera node may hang due to improper mutex handling: a thread held lock_sys.wait_mutex while triggering a streaming replication rollback, which also tried to acquire THD::LOCK_thd_kill, leading to incorrect mutex usage. In debug versions, this leads to diagnostics like "safe_mutex: Found wrong usage of mutex 'wait_mutex' and 'LOCK_thd_data'", but in both debug and release versions, there is some probability that the node may hang. (MDEV-36509)
corruption when query cache cannot allocate block (MDEV-34075)
Stack looping and SIGSEGV in Item_args::walk_args on UPDATE (MDEV-31647)
Server crash in find_field_in_tables, Assertion `name' failed in find_field_in_table_ref (MDEV-25012)
Long server_audit_file_path causes buffer overflow (MDEV-36245)
Server crash when inserting from derived table containing insert target table (MDEV-32086)
ST_PointFromWKB ignores SRID argument and always creates the POINT with 0 for it's SRID (MDEV-32619)
mariadb-dump used wrong quoting character (MDEV-36268)
After a corrupted table on one node triggers the cluster to vote to evict a node that failed a transaction, the current master can’t commit any more and hangs. (MDEV-34998, MENT-2081)
MDEV-32157 intended to fix spider wrapper so that it is case insensitive, among other things. However that fix was incomplete, as the udf spider_direct_sql may still require case sensitivity. MDEV-35807 fixes this. (MDEV-35807)
Creating partitioned tables is disallowed when wsrep_osu_method=TOI and wsrep_strict_ddl=ON, preventing alteration or deletion of partitioned tables. (MDEV-27861)
Attempting to create a CONNECT engine table results in "non-InnoDB sequences in Galera cluster" error message in logs due to an incorrect engine check. (MDEV-35748)
innodb_snapshot_isolation=1 gives error for not committed row changes (MDEV-36639)
MariaDB Backup returns with an error like "Error on file ./test/t1#P#p513.MYD open during `test`.`t1` table copy for partitioned MyISAM tables when running out of file handles (MENT-2089)
User without any privileges to a sequence can read from it and modify it via column default (MDEV-36413)
User has unauthorized access to a sequence through a view with security invoker (MDEV-36380)
group by handler missing constant fields when selecting from a view (MDEV-36307)
Tests calling the udf spider_copy_tables fail with --view-protocol (MDEV-36335)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Oracle Linux 8 (x86_64, ARM64)
Oracle Linux 9 (x86_64, ARM64)
Red Hat Enterprise Linux 8 (x86_64, ARM64)
Red Hat Enterprise Linux 9 (x86_64, ARM64, PPC64LE)
Rocky Linux 8 (x86_64, ARM64)
Rocky Linux 9 (x86_64, ARM64)
SUSE Linux Enterprise Server 12 (x86_64)
SUSE Linux Enterprise Server 15 (x86_64, ARM64)
Ubuntu 20.04 (x86_64, ARM64)
Ubuntu 22.04 (x86_64, ARM64)
Ubuntu 24.04 (x86_64, ARM64)
Red Hat UBI 8 (x86_64, ARM64)
Red Hat UBI 8 is part of the Enterprise Server Docker Image. It does not support MariaDB Enterprise Cluster (Galera) or MariaDB ColumnStore.
Server crashes when setting wsrep_cluster_address after adding invalid value to wsrep_allowlist table (MDEV-31173)
mariadbd hangs on startup when --init-file target does not exist (MDEV-34814)
Possible server crash in cases where the function DEFAULT() is part of a query (MDEV-35276)
Possible unique hash index corruption for system-versioned tables after DML including DELETE HISTORY statement. This can lead to data corruption, crashes, and incorrect query results (MDEV-33470)
Possible error log entry
This ALTER only fails if the type of field is changed in a multi-ALTER statement, where types of other fields are changed with the ALTER
Creation of a view with UNION and SELECT ... FOR UPDATE in the definition fails with error "ER_PARSE_ERROR (1064): You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near..." (MDEV-29537)
INSERT...SELECT on MyISAM or ARIA tables are replicated by MariaDB Enterprise Cluster (Galera) (MDEV-34647)
SELECT MIN on Spider table returns more rows than expected (MDEV-26345)
Can't selectively restore sequences using innodb tables from backup (MDEV-32350)
LOAD DATA INFILE with geometry data fails (MDEV-34883)
Executing an UPDATE statement in prepared statement mode having positional parameter bound with an array can result in an incorrect number of updated rows in case there is a BEFORE UPDATE trigger that executes yet another UPDATE statement to change a table (MDEV-34718)
Recovery fails to note some log corruption, resulting in "log sequence number in the future" error messages, and possibly adds more corruption (MDEV-34802)
When log corruption is noted, the server can now only be started when using the option innodb_force_recovery
Unexpected error "Row size too large (> 8123)..." for a correct INSERT after a instantly dropped BLOB column (MDEV-35122)
Wrong binlog timestamps on secondary nodes of MariaDB Enterprise Cluster (MENT-2164)
When the MariaDB Enterprise Audit Plugin is configured to log ACL queries, a statement CREATE USER .. IDENTIFIED VIA ed25519 USING PASSWORD(...) is logged including the password (MENT-2181)
When the MariaDB Enterprise Audit Plugin is configured to log ACL queries, passwords are masked for CREATE USER .. IDENTIFIED BY, but not for CREATE OR REPLACE USER, or SET STATEMENT ... FOR CREATE USER (MENT-2188)
Sequence Fails with Galera Version 26.4.16 and later in the Presence of enabled streaming replication (MDEV-34124)
The HashiCorp Vault key management plugin has been updated to improve availability during intermittent network or server issues. Previously, the plugin only allowed the use of cached keys if a connection timed out. This behavior has been broadened and made the default to prevent MariaDB Server interruptions. (MDEV-38203)
Expanded Error Handling: The plugin now treats common connectivity issues—such as “connection refused,” “no route to host,” and “internal server error”—the same as a timeout. This ensures the server continues to function using cached keys if the Vault server is briefly unreachable.
Updated Defaults:
The option to use cached values during communication errors is now enabled by default.
The cache timeout is now defaulting to the maximum value.
Audit log plugin will now log events for MariaDB Enterprise Cluster (Galera) replication applier operations with a generic user name: <wsrep_applier>. Before this change the user name in audit logs has been empty ()
Issues Fixed
Can result in data loss
Server crash upon moving InnoDB table with fulltext index between databases (MDEV-31892)
When innodb_buffer_pool_size is not large enough to buffer all log records during crash recovery, InnoDB may fail to recover a page from the doublewrite buffer. (MDEV-37558)
When an inplace ALTER operation is rolled back, InnoDB drops intermediate tables and their associated FTS internal tables. However, MariaBackup's DDL tracking can incorrectly report
An issue is fixed which could result in server hangs when a replica internally scans its mysql.gtid_slave_pos table contents. (MDEV-20586)
Fixed crash in Aria when doing a sub transaction like reading an entry from the proc table. (MDEV-23132)
Wrong variable name in the configuration leads Galera to think SST/IST failed, at next restart will request a full SST (MDEV-31517)
An unresolvable outer reference can cause a server to crash. ()
Segfault when preparing unreferenced select in recursive CTE ()
Server can crash when JSON_SCHEMA_VALID() is used ()
With DEFAULT() in the WHERE clause and with nested views, the server can crash ()
innodb_undo_log_truncate=ON leads to out-of-bounds write ()
When using UPDATE in a BEFORE trigger, the server can crash ()
Crash with low @@optimizer_max_sel_arg_weight and optimizer_trace=on. ()
InnoDB could crash during a workload that is frequently rebuilding or dropping tables by executing statements such as TRUNCATE TABLE, OPTIMIZE TABLE, or DROP TABLE. ()
Crash on reading corrupted view frm file ()
Galera cluster crashes when granting permission to non existing user after setting max_error_count and wsrep_ignore_apply_errors to zero ()
crash when reading a corrupted view frm file ()
Server crashes after INSERT.. ON duplicate KEY UPDATE i = DEFAULT; ()
Sever shutdown prevented by RocksDB setting rocksdb_pause_background_work ()
Secondary indexes could be corrupted for InnoDB tables that contain indexed virtual columns. ()
A large number of joins in a SELECT can crash the server ()
Specially crafted packet could cause the server to crash ()
The server can hang when data-at-rest encryption is used with multiple encryption threads ()
JSON_SCHEMA_VALID() crashes with a long enum ()
The authentication plugin caching_sha2_password can crash on large packets ()
Spider could crash when loop detection variables were artificially manipulated. ()
malformed spatial data could've caused a crash inside InnoDB ()
Long blob prefix keys could cause a crash in galera. ()
SHOW FUNCTION CODE package_func crashes on the second execution ()
Server crashes in setup_returning_fields upon 2nd execution of multi-table-styled DELETE from a view ()
With --encrypt-binlog=ON if a node fails to apply a writeset it will crash the whole cluster due to a bug in Galera library encryption handling. Fixed in Galera library ()
Can result in unexpected behaviour
JSON_UNQUOTE returns incorrect results depending on the collation (MDEV-14301)
Wrong warning (Errcode: 2 "No such file or directory") shown by SHOW WARNING when a partitioned table is changed with ALTER. Only "ALGORITHM=NOCOPY is not supported for this operation. Try ALGORITHM=INPLACE" should be shown (MDEV-35562)
A default value cannot be removed with ALTER and DROP DEFAULT default if the column is defined with NOT NULL ()
Bogus [ERROR] InnoDB: Compressed page checksum mismatch could be reported for ROW_FORMAT=COMPRESSED tables. ()
ed25519 client authentication plugin would fail to load for anything but mariadb client utility ()
The CHECK TABLE command can be executed by users with the table-level CREATE but not with a global CREATE privilege ()
sys.ps_setup_save does not restore SQL_LOG_BIN after error, causes replication discrepancy ()
If the server was killed during an operation that creates a .ibd file, such as TRUNCATE TABLE, it could fail to recover. ()
(Column) CHECK constraints can cause CREATE TABLE (SELECT) queries to fail ()
If the server was killed during an operation that creates a .ibd file, such as TRUNCATE TABLE, it could fail to recover. ()
event scheduler spams the error log ()
wrong result with index_merge_sort_intersection and rowid_filter=on ()
Auth Switch with Long Password Corrupts Database Name ()
Unexpected Data too long error on subselect as a multiplication argument ()
MariaDB sometimes reports "index is corrupted" for encrypted Aria tables, but this appears to be a false positive caused by an off-by-one check in the encryption post-read hook ()
Related to Install and Upgrade
In case if the view mysql.user was created (e.g. in 10.6) with a server with an unexpected character_set_client or collation_connection, e.g. utf8mb3 and utf8mb3_general_ci, mysql_upgrade does not fix it to the expected latin1 and latin1_swedish_ci. (MDEV-38698)
In case if the view mysql.user was created (e.g. in 10.6) with a server with an unexpected character_set_client or collation_connection, e.g. utf8mb3 and utf8mb3_general_ci, mysql_upgrade does not fix it to the expected latin1 and latin1_swedish_ci. (MENT-2523)
Related to performance
ASCII strings will be sent to client connections slightly faster. (MDEV-21816)
In multi-threaded write-heavy workloads, buf_pool.flush_list_mutex could become a point of contention. (MDEV-38069)
Poor cost calculations for index access cause bad query plans for big VARCHARs (MDEV-38164)
Selectivity sampling for LIKE %foo% not done when the table has no sargable conditions ()
Clustered PK not used for Loose Index Scan optimization ()
Under some circumstances, some read operations could unnecessarily wait for an InnoDB log write. ()
Galera threads time out under heavy load ()
Unexpected results
Unexpected success and result set upon erroneous JSON_VALUE call (MDEV-25148)
On an error in CREATE TEMPORARY TABLE ... SELECT, InnoDB had internally aborted a transaction, while the server thought the transaction still existed. This caused various inconsistencies in binlog, disappeared savepoints, etc. This was fixed differently in different versions. In 10.11.16, 11.4.10, 11.8.6 the transaction is consistently and completely rolled back if CREATE TEMPORARY TABLE ... SELECT fails with an error. In 12.3.1 there is no rollback at all. (MDEV-36787)
prohibited semijoin with rownum correctly process in HAVING rownum. ()
Incorrect results for INTERSECT ALL in ORACLE mode ()
Allow users with non-SELECT privileges to see rows in INFORMATION_SCHEMA.TRIGGERS table. ACTION_CONDITION, ACTION_STATEMENT and DEFINER columns are NULL unless the user is the owner of the schema or has TRIGGER privilege. ()
An empty value inserted for an ENUM column if a BEFORE trigger is used ()
Allow execution of multi-update/multi-delete queries within the storage engines. ()
unexpected type changing after changing AVG to MAX ()
INSERT ... RETURNING exposes columns for which the user lacks SELECT privilege ()
Presence of Spider disables "external" XA commit ()
Inconsistent behaviors when casting into time ()
Userstat Plugin CPU_TIME much higher than BUSY_TIME ()
Inconsistent results for make_set of cast ()
Potentially incorrect result with having reference to group by item. ()
An issue is fixed where GRANT EXECUTE ON PROCEDURE would fail on the master, but would still be replicated and executed successfully on the slave. ()
Changelog
For the complete list of changes in this release, see the changelog.
Platforms
In alignment to the enterprise lifecycle, MariaDB Enterprise Server 11.4.10-7 is provided for:
AlmaLinux 8 (x86_64, ARM64)
AlmaLinux 9 (x86_64, ARM64)
AlmaLinux 10 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Debian 12 (x86_64, ARM64)
Oracle Linux 8 (x86_64, ARM64)
Oracle Linux 9 (x86_64, ARM64)
Oracle Linux 10 (x86_64, ARM64)
Red Hat Enterprise Linux 8 (x86_64, ARM64)
Red Hat Enterprise Linux 9 (x86_64, ARM64, PPC64LE)
Red Hat Enterprise Linux 10 (x86_64, ARM64)
Rocky Linux 8 (x86_64, ARM64)
Rocky Linux 9 (x86_64, ARM64)
Rocky Linux 10 (x86_64, ARM64)
SUSE Linux Enterprise Server 15 (x86_64, ARM64)
Ubuntu 22.04 (x86_64, ARM64)
Ubuntu 24.04 (x86_64, ARM64)
Microsoft Windows (x86_64) (Without MariaDB Enterprise Cluster (Galera) support)
Red Hat UBI 8 (x86_64, ARM64)
Red Hat UBI 8 is part of the Enterprise Server Docker Image. It does not support MariaDB Enterprise Cluster (Galera) or MariaDB ColumnStore.
Some components of MariaDB Enterprise Server are supported on a subset of platforms. See MariaDB Engineering Policies for details.
MariaDB Enterprise Server 10.6.18-14 is a Stable (GA) maintenance release of MariaDB Enterprise Server 10.6.
MariaDB Enterprise Server 10.6.18-14 was released on 2024-06-11.
Fixed Security Vulnerabilities
CVE (with link)
CVSS base score
4.9
Notable Changes
Galera has been updated to 26.4.18
The GCS protocol version has been changed, preventing a downgrade of individual nodes of a MariaDB Enterprise Cluster
Changes to the use of the variable optimizer_adjust_secondary_index_costs (MDEV-33306)
The default of the variable optimizer_adjust_secondary_index_costs changed from "0" to "" and is now of type ENUM
The variable can now be set to a combination of the following values separated by a ',':
adjust_secondary_key_cost: Update secondary key costs for ranges to be at least 5x of clustered primary key costs.
Changes in Storage Engines
This release originally incorporated MariaDB ColumnStore engine version 23.10.1.
This release now incorporates MariaDB ColumnStore engine version 23.10.2.
Issues Fixed
Can result in data loss
With --gtid-ignore-duplicate set a transaction can be double-applied from another replication source if at applying the transaction from the initial source required retrying in parallel execution. (MDEV-33475)
Backups of server with innodb_encrypt_tables=1 can become corrupted in mariadb-backup --prepare (MDEV-33334)
With multiple GTID domains, optimistic parallel slave conflicts involving XA-PREPARE event group could not be resolved correctly which might cause unnecessary stop of the slave serve. ()
The TIMESTAMP value of '1970-01-01 00:00:00' can be inserted via a INSERT ...FROM ... SELECT in strict mode although it should result in an error ()
Galera-replicated events can in some cases contain the wrong time when versioning is used ()
wrong row targeted with insert ... on duplicate and replace, leading to data corruption ()
Can result in hang or crash
Using current MariaDB Enterprise Backup against an older server can result in a crash, as the system variable @@aria_log_dir_path does not exist (MDEV-31251)
When using a non-default setting for innodb_change_buffering, the server can crash (MDEV-33332)
When a fulltext search query with more than 4G inserted or updated rows is executed, the server can crash (MDEV-33383)
InnoDB is holding shared dict_sys.latch while waiting for FOREIGN KEY child table lock on DDL ()
InnoDB may hang when temporarily running out of buffer pool ()
For encrypted table spaces an ALTER operation can hang when an encryption thread works on the same tablespace ()
EXPLAIN statement that uses a subquery which has a query plan that A) will examine less than @@expensive_subquery_limit rows and B) will use join buffer could cause a crash. ()
For a SPIDER table, when deleting partitions from a table, the server can crash ()
After an unsuccessful ALTER TABLE on an ARIA table due to a full temp space, any subsequent query results in the following error if it involves using temp: ()
When replaying a binary log with mariadb-binlog, the tool can crash if the binary log includes statements related to sequences, like SELECT NEXTVAL(s) (MDEV-31779)
MariaDB Enterprise Backup fails with the following error message if the prepare step of the backup encounters a data directory which happens to store wsrep xid position in TRX SYS page: (MDEV-33540)
Assertion failure when a client disconnected during transaction that is in the XA PREPARE state or when incomplete transaction was recovered from undo logs on server startup and not yet rolled back (MDEV-33278)
Failure with the following error message when streaming transaction in idle state is BF aborted: (MDEV-33509)
Assertion when user commits an empty transaction for DDL that was killed during certification: (MDEV-32787)
MDL lock conflict can occur if the transactions acquire explicit MDL locks from the InnoDB level when persistent statistics is re-read for a table and such a transaction is subject to BF-abort (MDEV-33136)
Graceful node shutdown can crash garbd and MariaDB Enterprise Cluster can go non-primary when SSL is used (MDEV-33495)
Can result in unexpected behavior
Wrong result sets are returned by the second execution of prepared statements from selects using mergeable derived tables pushed into external engine (MDEV-31361)
IMPORT TABLESPACE fails with column count or index count mismatch: (MDEV-30655)
A query that uses a derived table that employs constructs with side-effects, like (SELECT @var:=... ) as derived_tbl, could produce wrong results (MDEV-30975)
ORDER BY COLLATE improperly applied to non-character columns which is resulting in an unordered result set (MDEV-33318)
Inconsistent behaviors of UPDATE under Read Uncommitted & Read Committed isolation level (MDEV-26643)
When two transactions modify the data at the same time with isolation level REPEATABLE-READ, the latter transaction does take the change from the first transaction into account ()
Spider/ODBC passed double quotes for names, in ANSI style (MENT-958)
Default charset doesn't work with PHP MySQLi extension ()
Bad SEPARATOR value in GROUP_CONCAT on character set conversion can lead to a wrong result ()
Spider returns parsing failure on valid left join select by translating the on expression to () ()
When creating a temporary InnoDB table with CREATE TEMPORARY TABLE ... AS SELECT ... from an InnoDB table as a non SUPER / READ ONLY ADMIN user, the following error is shown instead of creating the table: ()
CREATE TEMPORARY TABLE (without SELECT), INSERT ... SELECT, and CREATE ... LIKE are not affected by this bug
Phantom rows caused by UPDATE of PRIMARY KEY (MDEV-32898)
Mariadb-dump trusts the server and does not validate the data. A modified dump file can include system commands used by the mariadb-client. Dumps are now loaded in the sandbox mode by default, a system call will result in an error ()
Updating a case insensitive large unique key with an insensitive change of the value can result in a duplicate key error ()
When setting binlog_annotate_row_events=1, an event of binlog file can be truncated ()
Wrong result with semi-join and split-table derived table from queries with an IN subquery predicate in the WHERE clause and a derived table in the FROM clause to which split materialized optimization could be applied. ()
With galera, correct transactions could not be committed with the following error when accessing system tables for read, and write to innodb tables in the same transaction: ()
A wrong result on 2-nd execution of a prepared statement is possible when selecting from a view using a merged derived table (MDEV-31277)
Original IP not shown in network related error messages when proxy_protocol is in use (MDEV-33506)
Incorrect DEFAULT expression evaluated in UPDATE (MDEV-33790)
group by optimization incorrectly removing subquery where subject buried in a function ()
Related to performance
Replication with XA events can show decreased performance. Adapt parallel slave's round-robin scheduling to XA events (MDEV-33668)
Row locks for non-modified rows are not released at XA PREPARE (MDEV-33454)
Optimizer is sometimes choosing an index for queries with GROUP BY when it shouldn't, resulting in decreased performance. To protect compatibility to the current behavior @@optimizer_adjust_secondary_key_costs="disable_forced_index_in_group_by" has to be set to enable the fix ()
Table is getting rebuild with ALTER TABLE ADD COLUMN although it should be an instant operation not requiring a rebuild ()
Semi-sync Wait Point AFTER_COMMIT Slow on Workloads with Heavy Concurrency ()
Aggregation functions fail to leverage uniqueness property ()
Generally, computing aggregate function with DISTINCT argument: aggregate_func(DISTINCT col1, col2, ...) requires producing a de-duplicated set of its arguments, which can be CPU-intensive
When we select from one table the argument list includes the table's PRIMARY (or UNIQUE) key:
then the arguments are guaranteed not to have duplicates. Such cases are now detected allowing the optimizer to skip de-duplication.
Changelog
For the complete list of changes in this release, see the changelog.
Platforms
In alignment to the enterprise lifecycle, MariaDB Enterprise Server 10.6.18-14 is provided for:
AlmaLinux 8 (x86_64, ARM64)
AlmaLinux 9 (x86_64, ARM64)
CentOS 7 (x86_64)
Debian 10 (x86_64)
Debian 11 (x86_64, ARM64)
Debian 12 (x86_64, ARM64)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Red Hat Enterprise Linux 7 (x86_64)
Red Hat Enterprise Linux 8 (x86_64, ARM64)
Red Hat Enterprise Linux 9 (x86_64, ARM64, PPC64LE)
MariaDB Enterprise Server 10.6.20-16 is a Stable (GA) maintenance release of MariaDB Enterprise Server 10.6.
MariaDB Enterprise Server 10.6.20-16 was released on 2024-12-10.
Backports
New, Detailed Replication Lag Representation (MENT-2095)
The Seconds_Behind_Master field of SHOW REPLICA STATUS can be complex and confusing, especially when parallel replication, delayed replication, or the option sql_slave_skip_counter is used. To help provide a consistent view of replication lag, three new fields have been added to the statement's output to provide specific timing information about the state of the IO and SQL threads.
Three new values have been added to the replica status:
Master_last_event_time
Timestamp of the last event read from the primary by the IO thread
Slave_last_event_time
Timestamp from the primary of the last event committed on the replica
Master_Slave_time_diff
The difference of the above two timestamps
New Information Schema Table For Password Related Data (MENT-2145)
A new information Schema view, USERS, has been added, which DBAs can use to get insights about password related information for a user. This information can be used:
by an application to inform a user about a password about to expire or an account which is at risk of being blocked due to the number of wrong passwords entered
by DBAs to query users which have been blocked because of too many invalid passwords entered
GTID binlog events now include the thread ID (MENT-2180)
The thread ID and the corresponding statement can now be retrieved from binary logs
The output of mariadb-binlog also includes the thread ID
Notable Changes
A new parameter --quick-max-column-width is now available in the mariadb client to limit the field width when used with the --quick mode (MDEV-34704)
JSON_TABLE allows to specify default column values, but they had to be specified as string constants (MDEV-25822)
With this patch, one can specify default values as any literals (string, integer, decimal number, date literal, etc)
Example:
A new option --skip-freed-pages has been added to the innochecksum tool (MENT-2183)
This option can be used to skip reporting freed undo logs as existing undo log pages
Possible corruption due to a rare race condition between page creation and eviction (MDEV-34453)
An error like "[ERROR] InnoDB: Trying to read ... bytes at ... outside the bounds of the file" will be shown in the error log
Using group_concat with a compressed or GIS column can lead to a server crash and potential data corruption for users performing group_concat operations on tables containing these data types ()
Unique hash index corruption for system-versioned tables after DML including a DELETE HISTORY statement, causing data corruption and error 'Record in index not found on update' ()
Can result in hang or crash
When using Total Order Isolation (TOI) executing ALTER TABLE while a SR transaction is in progress, the user encounters a freeze and a server error "InnoDB: WSREP: BF lock wait long for trx" (MDEV-34836)
A query could cause a crash if it has a HAVING clause with a construct tblX.column=column_or_constant and the optimizer was able to infer that table tblX is a constant table. Note that the HAVING clause may be from the original query or may come from ) (MDEV-23983)
Calling a UDF for the engine SPIDER results in a crash if the SPIDER storage engine plugin could not be loaded (MDEV-34682)
The use of XA PREPARE may cause InnoDB to hang ()
Under Windows Subsystem for Linux, InnoDB crashes on ALTER TABLE or OPTIMIZE TABLE ()
Server crashes when setting wsrep_cluster_address after adding invalid value to wsrep_allowlist table ()
mariadbd hangs on startup when --init-file target does not exist ()
Possible server crash in cases where the function DEFAULT() is part of a query ()
Primary stalling when a DDL is executed using semi-sync replication with wait point = AFTER_SYNC (MENT-2162)
Deadlock on Replica during BACKUP STAGE BLOCK_COMMIT on XA transactions (MENT-2152)
In some situations, where a MariaDB Enterprise cluster (Galera) node crashes, some threads are still working. This is visible via the error log, which is still getting errors written by the node. As the node stops interacting correctly with other cluster nodes and is blocking them from taking over the primary state, the whole cluster hangs ()
When switching to the SQL Mode EXTENDED_ALIASES and selecting from a table of type SPIDER, the server can crash (MENT-2108)
Can result in unexpected behavior
When a user runs mariadb-binlog with --stop-position, they would expect the output to contain events up to that event. If the output did not contain events up to that event, this may result in various unexpected behaviors, e.g., an incomplete database state if they piped the output to the mariadb client and expected certain transactions to have executed in the database, yet never were run (MDEV-27037)
A table name can not be reused when the table was changed using "ALTER TABLE...STATS_PERSISTENT=0" and was dropped afterwards. This also results in the error "Can't write; duplicate key in table 'mysql.innodb_table_stats" in tsing ALTER TABLE...STATS_PERSISTENT=0 fails to drop statistics" (MDEV-34207)
Changing a data type of a field used in a foreign key constraint fails with Error "Cannot change column '...': used in a foreign key constraint '...'" ()
This ALTER only fails if the type of field is changed in a multi-ALTER statement, where types of other fields are changed with the ALTER
Creation of a view with UNION and SELECT ... FOR UPDATE in the definition fails with error "ER_PARSE_ERROR (1064): You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near..." ()
In some contexts text strings containing numbers with the integer part equal to zero and the fractional part not equal to zero were erroneously evaluated as FALSE in WHERE condition ()
INSERT...SELECT on MyISAM or ARIA tables are replicated by MariaDB Enterprise Cluster (Galera) ()
Storage Engine S3 causes 500 error when using Huawai Cloud S3 and possibly other S3 providers ()
A new option S3-provider has been added. S3-provider=Huawai needs to be set for Huawai Cloud S3
JSON_TABLE doesn't properly unquote strings ()
SELECT MIN on Spider table returns more rows than expected ()
Can't selectively restore sequences using innodb tables from backup ()
LOAD DATA INFILE with geometry data fails ()
Executing an UPDATE statement in prepared statement mode having positional parameter bound with an array can result in an incorrect number of updated rows in case there is a BEFORE UPDATE trigger that executes yet another UPDATE statement to change a table ()
Recovery fails to note some log corruption, resulting in "log sequence number in the future" error messages, and possibly adds more corruption ()
When log corruption is noted, the server can now only be started when using the option innodb_force_recovery
When SET GLOBAL innodb_adaptive_hash_index=ON is set, ALTER TABLE...IMPORT TABLESPACE with FULLTEXT SEARCH may corrupt the adaptive hash index ()
InnoDB fails to merge the change buffer to ROW_FORMAT=COMPRESSED tables. This can result in secondary indexes in ROW_FORMAT=COMPRESSED tables becoming corrupted, which in turn would lead to wrong results for queries that use those secondary indexes ()
CHECK TABLE would notice the corruption
OPTIMIZE TABLE would fix the corruption
No error/warning returned when a transaction is executed, which includes transactional and non-transactional engines ()
When using the experimental option wsrep_mode=REPLICATE_ARIA a transaction can possibly include changes to tables using transactional and non-transactional engines. But only the changes of the transactional engine will be rolled back in case of a conflict
NOTE: Experimental features should not be used on production
Unexpected error "Row size too large (> 8123)..." for a correct INSERT after a instantly dropped BLOB column ()
Wrong binlog timestamps on secondary nodes of MariaDB Enterprise Cluster (MENT-2164)
InnoDB fails to shrink the system tablespace when it contains leaked undo log pages. It fails to free the unused segment if a XA PREPARE transaction exist or if the previous shutdown was not done with innodb_fast_shutdown=0 (MENT-2148)
The :autoshrink attribute needs to be set for the system tablespace to free disk space on startup
When the MariaDB Enterprise Audit Plugin is configured to log ACL queries, a statement CREATE USER .. IDENTIFIED VIA ed25519 USING PASSWORD(...) is logged including the password (MENT-2181)
When the MariaDB Enterprise Audit Plugin is configured to log ACL queries, passwords are masked for CREATE USER .. IDENTIFIED BY, but not for CREATE OR REPLACE USER, or SET STATEMENT ... FOR CREATE USER (MENT-2188)
Related to performance
Bug that causes blocking issues when other transactions attempt to acquire locks on records held by an XA transaction in the prepared state, even if those records haven't been modified (MDEV-34466)
This affects InnoDB storage engine, XA transactions, and non-blocking mode
Unnecessary copying of log records by MariaDB Enterprise Backup when further transaction commits are blocked by BACKUP STAGE BLOCK_COMMIT (MENT-2133)
It also would cause further effort of rolling back incomplete transactions after the backup is restored
Aria internal temporary tables unnecessarily write all changed blocks to disk when the table is closed at end of query (MENT-2182)
Changelog
For the complete list of changes in this release, see the changelog.
Platforms
In alignment to the enterprise lifecycle, MariaDB Enterprise Server 10.6.20-16 is provided for:
AlmaLinux 8 (x86_64, ARM64)
AlmaLinux 9 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Debian 12 (x86_64, ARM64)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Red Hat Enterprise Linux 8 (x86_64, ARM64)
Red Hat Enterprise Linux 9 (x86_64, ARM64, PPC64LE)
MariaDB Enterprise Server 10.6.5-2 is a Stable (GA) maintenance release of 10.6. This release includes a variety of fixes.
MariaDB Enterprise Server 10.6.5-2 was released on 2021-12-13.
Fixed Security Vulnerabilities
Notable Changes
Galera updated to 26.4.10.
Debian 11 support added.
no longer enables pushdown of UDFs and Stored Functions to the Data Node by default: ()
Issues Fixed
Can result in data loss
InnoDB tables with ROW_FORMAT=COMPRESSED could be corrupted. (MENT-1367)
Can result in a hang or crash
, powered by Galera, can crash on if the table does not have a primary key and if the data for a field exceeds 4096 bytes. ()
When an InnoDB tablespace (.ibd) file is imported using without a corresponding .cfg file, InnoDB causes a server crash. (, )
When (or ) is executed against an InnoDB table with a FULLTEXT
Can result in unexpected behavior
skip_networking does not prevent replication. ()
joiner node incorrectly uses localhost for TLS certificate verification and fails to join cluster when and encrypt=3 are configured. ()
Related to install and upgrade
On CentOS 7, auth_pam_tool has incorrect permissions. ()
Running with no 'root'@'localhost' user may fail with the message The user specified as a definer ('root'@'localhost') does not exist FATAL ERROR: Upgrade failed. ()
Changes in Storage Engines
This release originally incorporated MariaDB .
This release now incorporates .
Platforms
In alignment to the , MariaDB Enterprise Server 10.6.5-2 is provided for:
CentOS 7 (x86_64)
Debian 9 (x86_64, ARM64)
Debian 10 (x86_64, ARM64)
Debian 11 (x86_64, ARM64)
Some components of MariaDB Enterprise Server might not support all platforms. For additional information, see
Release Notes for MariaDB Enterprise Server 10.4.33-23
MariaDB Enterprise Server 10.4.33-23 is a maintenance release of . This release includes a variety of fixes.
MariaDB Enterprise Server 10.4.33-23 was released on 2024-03-11.
Notable Changes
Galera has been updated to 26.4.17
Issues Fixed
Can result in data loss
BLOB corruption on UPDATE of PRIMARY KEY with FOREIGN KEY ()
Corrupted table after instant ADD COLUMN failed due to attempting to exceed the maximum record length ()
Can result in hang or crash
Crash while trying to complain "unsupported meta-data version" on ALTER TABLE...IMPORT TABLESPACE with MySQL 8.0 files ()
Wrong table attribute PAGE_COMPRESSED=1 shown for tables in the InnoDB system tablespace ()
MariaDB crashes with foreign_key_checks=0 when changing a column and adding a foreign key at the same time ()
and
GTIDs may diverge in Galera cluster after CREATE TABLE AS ()
This can subsequently lead to diagnostics like:
and node crash:
SHOW SLAVE STATUS can deadlock an errored replica when a parallel replica worker gets killed at about when SHOW is issued ()
When executing SELECT BINLOG_GTID_POS(@binlog_file...) with the user variable being NULL, the server can crash ()
A Query that uses a specific SQL construct can cause a server crash. The construct is an equality comparison of table-less row subquery and a subquery that has a UNION operation at the top level: (
Can result in unexpected behavior
SHOW WARNINGS can show wrong foreign key related warnings/errors from an earlier transaction for InnoDB ()
LeakSanitizer errors in mem_heap_create_block_func upon query from I_S.INNODB_SYS_TABLES with LIMIT ROWS EXAMINED ()
Related to performance
An ALTER using ALGORITHM=INPLACE or adding new indexes can increase the table space file size ()
Operations that involve extending InnoDB files can be extremely slow when the data directory resides on an NFS server that uses a smaller block size than 4096 bytes ()
* Platforms
In alignment to the , MariaDB Enterprise Server 10.4.33-23 is provided for:
CentOS 7 (x86_64)
Debian 10 (x86_64, ARM64)
Microsoft Windows (x86_64) (MariaDB Enterprise Cluster excluded)
Red Hat Enterprise Linux 7 (x86_64)
Some components of MariaDB Enterprise Server are supported on a subset of platforms. See for details.
ERROR 1808 (HY000): Schema mismatch (Number of columns don't match, table has x columns but the tablespace meta-data file has y columns)
ERROR 1290 (HY000): The MariaDB server is running with the --read-only option so it cannot execute this statement
Transactional commit not supported by involved engine
SELECT aggregate_func(DISTINCT t1.primary_key, ...) FROM t1;
SELECT *
FROM
json_table('{"a": "123"}',
'$' columns(col1 INT PATH '$.a' DEFAULT 1 ON empty)) AS T;
SELECT *
FROM
json_table('{"a": "123"}',
'$' columns(col1 DATE PATH '$.DATE' DEFAULT DATE '2020-01-01' ON empty)) AS T;