Release date: 10 May 2023
MariaDB 10.6 is the current long-term series of MariaDB, maintained until July 2026. It is an evolution of MariaDB 10.5 with several entirely new features.
MariaDB 10.6.13 is a Stable (GA) release.
Thanks, and enjoy MariaDB!
Crash on in a table ()
UNIQUE USING HASH accepts duplicate entries for tricky collations ()
rec_get_offsets() is not optimal ()
Performance regression in fil_space_t::try_to_close() introduced in ()
mariadb-backup doesn't utilise innodb-undo-log-directory (if specified as a relative path) during copy-back operation ()
mariadb-backup issues error messages during InnoDB tablespaces export on partial backup preparing ()
mariadb-backup does not copy Aria logs if aria_log_dir_path is used ()
Race condition between buffer pool flush and log file deletion in mariadb-backup --prepare ()
Fixed a deadlock on parallel slave involving full image Write event on the sequence engine ()
Fixed an attempted out-of-order binlogging error on slave involving ALTER on the sequence engine ()
Corrected non-versioned master to versioned slave replication on no-unique attribute table ()
Mended encrypted binlog master to error out to gtid-mode slave when master could not decrypt a binlog file ()
optimization is improved to re-fill the materialized table only if necessary. The fewer number of table refills is taken into account when choosing query plan, too ().
New optimizer_switch option, , is added. It is off by default. When set to ON, the optimizer will produce tighter bounds for hash join output cardinality. ()
Queries using SELECT DISTINCT some_expression(aggregate_function()) could produce wrong query result. ()
Add replication setup to containers contributed by Md Sahil ()
As per the , this will be the last release of for Fedora 36.
In this release repositories for Fedora 38 and Ubuntu 23.04 Lunar have been added.
Fixes for the following :
For a complete list of changes and bugfixes made in , with links to detailed information on each push, see the .
For a full list of contributors to , see the .
InnoDB recovery hangs when buffer pool ran out of memory (MDEV-30551)
InnoDB undo log truncation fails to wait for purge of history (MDEV-30671
MariaDB crash due to DB_FAIL reported for a corrupted page (MDEV-30397)
Deadlock between INSERT and InnoDB non-persistent statistics update (MDEV-30638)
InnoDB hang on B-tree split or merge (MDEV-29835)
Performance regression in locking reads from secondary indexes (MDEV-30357)
Improve adaptive flushing (MDEV-26055)
Make page flushing even faster (MDEV-26827)
Purge misses a chance to free not-yet-reused undo pages (MDEV-29593)
InnoDB temporary tablespace: reclaiming of free space does not work (MDEV-26782)
Fix miscount of doublewrites by (MDEV-31124)
Refined optimistic parallel slave to error-exit without any hang (MDEV-30780)
Ensured SHOW-SLAVE-STATUS is processed on the parallel slave having a necessary mutex always intialized (MDEV-30620)
Fixed the slave applier to report a correct error when gtid_slave_pos insert fails for some (engine) reasons (MDEV-31038)
Made parallel slave reports in performance schema consistent with that of show-slave-status (MDEV-26071)
now prints more information about Block Nested Loop joins: block-nl-join element now has r_loops, r_effective_rows and r_other_time_ms fields (MDEV-30806, MDEV-30830, MDEV-30972).
A GROUP BY query with MIN(primary_key) in select list and primary_key<>const in the WHERE could produce wrong result when executed with "Using index for group-by" strategy (MDEV-30605)
EXPLAIN could erroneously report that is used for partitioned tables. Partitioned tables do not support it. (MDEV-30596)
A bug in selectivity computations for SINGLE/DOUBLE_PREC_HB histograms could cause wrong estimates to be produced. This could cause the optimizer to pick sub-optimal query plans (MDEV-31067).
This page is licensed: CC BY-SA / Gnu FDL