This is a milestone release, for use at your own risk. Upgrades between milestone releases (or from a milestone release to a GA release) are not supported. Significant development changes take place in milestone releases and you may encounter compatibility issues, such as data format changes that require attention in addition to the usual procedure of running mysql_upgrade. For example, you may find it necessary to dump your data with mysqldump before the upgrade and reload it afterward. (Making a backup before the upgrade is a prudent precaution in any case.)
During data directory initialization or upgrade, MySQL now creates a
'mysql.session'@'localhost'reserved account. This account is used internally by plugins to access the server. It is locked so that it cannot be used for client connections. (Bug #25642343)
These system variables now are available to define mandatory roles and to enable granted roles to be automatically activated at client connection time:
For more information, see Using Roles.
For Unicode data that uses
NO PADcollations, sorting of multibyte and variable-length values has been improved:
NO PADcollations are those based on UCA 9.0.0 and higher, such as
The performance improvement is greatest for key values with short weight strings; that is, strings where the weight strings do not fill their entire permitted length. For a
VARCHAR(10)column that uses the
utf8mb4_0900_ai_cicollation, values may take up to 160 bytes. The string
'a'uses only 2 bytes out of a possible 160 and is more sparse than
'abcdefghij'. But even
'abcdefghij'uses only 20 bytes of a possible 160 and is more sparse than a string of, for example, ten copies of U+337F SQUARE CORPORATION (㍿), whose weight string requires the entire 160 bytes available.
(Bug #25750527, Bug #85546)
MySQL now supports a new collation,
utf8mb4_0900_as_ci, for the
utf8mb4Unicode character set. This collation is accent sensitive and case insensitive. It is similar to the default
utf8mb4_0900_ai_ci) except that the default collation is accent insensitive.
MySQL also now supports a new Japanese collation,
utf8mb4_ja_0900_as_cs_ks, for the
utf8mb4Unicode character set. This collation is like
utf8mb4_ja_0900_as_csin that it is accent sensitive and case sensitive, but
utf8mb4_ja_0900_as_cs_ksis also kana sensitive and distinguishes Katakana characters from Hiragana characters.
utf8mb4_ja_0900_as_cstreats Katakana and Hiragana characters as equal for sorting. Applications that require a Japanese collation but not kana sensitivity may use
utf8mb4_ja_0900_as_csfor better sort performance.
utf8mb4_ja_0900_as_csuses three weight levels for sorting;
utf8mb4_ja_0900_as_cs_ksis the first collation to use the
_kscollation suffix. Japanese collations without this suffix are not kana sensitive.
For more information, see Unicode Character Sets.
These character set changes were made to MySQL client support:
These clients use a default character set of
latin1: mysql, mysql_upgrade, mysqladmin, mysqlcheck, mysqldump, mysqlimport, mysqlpump, mysqlslap, mysqlshow, and mysqltest. (This change does not affect character set autodetection performed by mysql, mysqladmin, mysqlcheck, mysqlimport, and
mysqlshow, as described at Connection Character Sets and Collations).
Client connections created using the
libmysqlclientlibrary use a default character set of
The mysqltest program supports a
--default-character-setoption for specifying the character set explicitly.
Performance: The strict aliasing optimization for GCC is no longer disabled, which results in a minor improvement for MySQL single-threaded performance.
Linux: MySQL now compiles on Alpine Linux. (Bug #25945568, Bug #80322)
Solaris: On Solaris, server builds now use
std::atomic, so they must be linked against
libstatomic. (Bug #25957991)
MySQL now compiles using Visual Studio 2017. (Bug #25788406)
The minimum version of the Boost library for server builds is now 1.64.0. (Bug #25772329, Bug #85593)
These changes were made with respect to client program development:
Client programs should only need to
<mysql.h>header file. In particular,
<my_config.h>should not be needed, and is no longer installed.
my_init()function is no longer included in the list of symbols exported from
libmysqlclient. It need not be called explicitly by client programs because it is called implicitly by other C API initialization functions.
For the Xcode IDE, header files are added to MySQL project sources so they can be searched. (Bug #25636986)
Work was done to clean up the source code base, including: Removing unneeded CMake checks; removing unused macros from source files; reorganizing header files to reduce the number of dependencies and make them more modular, removing function declarations without definitions, replacing locally written functions with equivalent functions from industry-standard libraries.
New services are available to enable server components to register and unregister system variables (
component_sys_variable_unregister) and to register status variables (
Two new services are available to enable server components and plugins to register and unregister user-defined-functions (UDFs):
mysql_service_udf_registration_aggregateprovide registration services for scalar and aggregate UDFs, respectively. These services enable components and plugins to manage UDFs for themselves, without the need for
mysql_stringstring-manipulation service has been converted from a plugin service to a component service: a set of string service APIs for use by server components. The password validation component has been updated to use the revised service implementation.
LOAD DATAoperations was enabled by default in some contexts: The client library in MySQL binary distributions was compiled with client-side
LOCALcapability enabled, and the
local_infilesystem variable was enabled on the server side.
LOCALcapability is now disabled by default in all contexts. Client programs must be configured explicitly to enable
LOCAL, and the server must be run with
local_infileenabled. See Security Considerations for LOAD DATA LOCAL. (Bug #24511108)
mysqld startup behavior has been modified as follows. In general, these changes enable more consistent or easier server startup when not using mysqld_safe.
The server executable determines its own full path name at startup and uses the parent of the directory in which it is located as the default
basedirvalue. This in turn enables the server to use that
basedirwhen searching for server-related information such as the
sharedirectory containing error messages.
If error log output cannot be redirected to the
--log-erroroption value, error output could be lost. This is now less likely.
If the server is started using the
--daemonizeoption and is not connected to a tty device, a default error logging option of
--log-error=""is used in the absence of an explicit logging option, to direct error output to the default log file. Previously, an error could occur under these circumstances.
(Bug #20398088, Bug #75343)
MySQL now supports a
SET PERSIST_ONLYvariant of
SETstatement syntax, for making configuration changes at runtime that also persist across server restarts. Like
SET PERSIST_ONLYwrites the variable setting to an option file named
mysqld-auto.cnfin the data directory. However, unlike
PERSIST_ONLYdoes not modify the runtime global system variable value. This makes
PERSIST_ONLYsuitable for configuring read-only system variables that only be set can at server startup. For more information, see Using Option Files, and SET Syntax for Variable Assignment.
explicit_defaults_for_timestampsystem variable is now enabled by default (previously disabled by default), and a warning occurs if you disable it. This means that the nonstandard (and deprecated) behaviors for default values and
NULL-value handling in
TIMESTAMPcolumns are now disabled by default.
column_statssystem table has been removed and replaced by the
column_statisticsdata dictionary table.
versiondata dictionary table was renamed to
INFORMATION_SCHEMAtables created by dynamic plugins now is recorded in the data dictionary and visible through the
These tables are no longer visible in
TABLES_DYNAMIC. User impact is minimal because these tables are for internal use by server components.
InnoDBstorage engine now uses the MySQL data dictionary rather than its own storage engine-specific data dictionary. For information about the data dictionary, see MySQL Data Dictionary.
The following list briefly describes the main implications of this change:
Upgrade and downgrade implications:
To upgrade from MySQL 5.7 to MySQL 8.0, you must perform the upgrade procedure described at Upgrading MySQL.
Downgrading from MySQL 8.0 to MySQL 5.7 is only supported using the logical downgrade method (a mysqldump downgrade). In-place downgrades are not supported.
Metadata updates associated with exporting and importing tablespaces using the transportable tablespace feature are now performed on global data dictionary tables instead of
InnoDBdata dictionary tables.
InnoDBin-memory metadata is now instantiated from global data dictionary objects. This metadata was previously read from
Table options that signify tablespace encryption and transparent page compression are now retrieved from the global data dictionary.
Data dictionary support was added for
FULLTEXTindexes. Auxiliary index table names were changed to lowercase.
InnoDBmetadata created or modified during DDL operations is now written to the global data dictionary.
mysqlsystem tables and data dictionary tables are now created in a single
InnoDBtablespace file named
mysql.ibdin the MySQL data directory. Previously, these tables were created in individual
InnoDBtablespace files in the
mysqldatabase directory. Associated changes include:
InnoDBdata dictionary buffer table, which stores fast changing
InnoDBmetadata, was renamed to
innodb_dynamic_metadataand moved from the
InnoDBsystem tablespace to the data dictionary tablespace.
Undo tablespace metadata now resides in the data dictionary tablespace.
Temporary tablespace metadata now resides in the data dictionary tablespace.
Table definitions for tables created in the
InnoDBsystem tablespace (
innodb_system), or general tablespaces now include a
TABLESPACEattribute. This change has the following implications:
You cannot use
CREATE TEMPORY TABLE ... LIKEto create an empty table based on the definition of a table that resides in any of the aforementioned tablespaces, as these tablespaces do not support temporary tables.
CREATE TABLE ... LIKEpreserves the
TABLESPACEattribute of the original table and creates a new table in the defined tablespace regardless of the
innodb_file_per_tablesetting. This is a temporary regression.
Comment stripping is now deprecated. This feature and the options to control it will be removed in a future MySQL version.
Support for these deprecated syntax constructs for table and column references has been removed and their use now results in an error. Instances of these constructs should be changed to remove the leading period.
Symbolic link support as described at Using Symbolic Links for MyISAM Tables on Unix, along with the
--symbolic-linksoption that controls it, is now deprecated and will be removed in a future MySQL version. In addition, the option is now disabled by default. The related
have_symlinksystem variable also is deprecated and will be removed in a future MySQL version.
MySQL now supports window functions that, for each row from a query, perform a calculation using rows related to that row. These include functions such as
NTILE(). In addition, most existing aggregate functions now can be used as window functions; for example,
AVG(). For more information, see Window Functions.Note
Each of the following words now is a reserved word and cannot be used as an identifier without identifier quoting:
Incompatible Change: These error-logging changes have been made:
The server is more forgiving if it cannot find the configured error-message file (specified using the
lc_messagessystem variables). Previously, the server wrote a message to the error log to indicate the problem, then aborted the startup process and exited. Now the server writes a message, but continues startup and defaults to built-in English messages. This applies to messages the server writes to the error log and sends to clients. See Setting the Error Message Language.
Error logging was rewritten to use the MySQL component architecture. Traditional error logging is implemented using built-in components, and logging using the system log is implemented as a loadable component. In addition, a loadable JSON-format log writer is available. To control which log components to enable, use the
log_error_servicessystem variable. For more information, see The Error Log.
Incompatibility: To enable logging to the system log, you must load the
log_sink_syseventloglog component and list it in the
log_error_servicesvalue (see Error Logging to the System Log). This differs from MySQL 5.7 and earlier, for which logging to the system log is enabled by default on Windows, and on all platforms requires no component loading.
A consequence of this configuration change is that the
log_syslogsystem variable previously used to control logging to the system log is obsolete and changes to its value have no effect.
log_syslogis now deprecated and will be removed in a future MySQL version.
MySQL now enables management of histogram statistics for table column values:
ANALYZE TABLEstatement supports
DROP HISTOGRAMclauses for generating and removing column histogram statistics.
The server stores histogram information in the
column_statisticsdata dictionary table. Histograms are viewable using the
histogram_generation_max_mem_sizesystem variable controls the amount of memory available for histogram generation.
The Performance Schema has a
memory/sql/histogramsinstrument for monitoring memory allocations performed for histogram generation.
--column-statisticsoption to add
ANALYZE TABLEstatements to the output to generate histogram statistics for dumped tables when the dump file is reloaded.
mysqladmin was added to Docker/Minimal packages because it is needed by InnoDB Cluster. (Bug #25998285)
For Windows, MSI installer packages now include a check for the required Visual Studio redistributable package, and produce a message asking the user to install it if it is missing. (Bug #25658832)
Debian/Ubuntu packages now support multiple MySQL instances with systemd. See Managing MySQL Server with systemd. (Bug #24559588, Bug #82785)
The parser rules for
ALTER TABLEwere refactored to be context independent and improve maintainability and extensibility. A resulting effect is that some previously accepted undocumented syntax variants are no longer accepted. For example,
CREATE TABLEstatements were permitted with column names qualified by the table name, or by the current database and table name. Such statements now produce an error.
The Performance Schema now supports versioning, and maintains the current definitions for its tables internally. At startup, the server compares its supported Performance Schema version with the Performance Schema version stored in the data dictionary. If the versions differ, the server drops any old Performance Schema tables and recreates them using the current definitions. In consequence of this change:
For MySQL upgrades, it is no longer necessary to run mysql_upgrade to incorporate changes to Performance Schema tables because they are recreated automatically as necessary at server startup.
mysql_system_tables.sqlsupport script no longer includes SQL statements for Performance Schema table creation because these table definitions are maintained internally.
To support dynamic Performance Schema table manipulation, a new component service named
pfs_table_serviceis now available.
The Performance Schema default instrumentation settings have changed:
transactioninstrument and the
events_transactions_historyconsumers are now enabled by default. See Performance Schema Transaction Tables.
memory/%memory instruments are now enabled by default. See Memory Summary Tables.
wait/lock/metadata/sql/mdlmetadata lock instrument is now enabled by default. See The metadata_locks Table.
The linked OpenSSL library for the MySQL Commercial Server has been updated to version 1.0.2l. Issues fixed in the new OpenSSL version are described at http://www.openssl.org/news/vulnerabilities.html.
This change does not affect the Oracle-produced MySQL Community build of MySQL Server, which uses the yaSSL library instead. (Bug #26160962)
Previously, any user could execute the
XA RECOVERstatement to discover the XID values for outstanding prepared XA transactions, possibly leading to commit or rollback of an XA transaction by a user other than the one who started it. Now
XA RECOVERis permitted only to users who have the new
XA_RECOVER_ADMINprivilege, which is expected to be granted only to administrative users who have need for it. This might be the case, for example, for administrators of an XA application if it has crashed and it is necessary to find outstanding transactions started by the application so they can be rolled back. This privilege requirement does not affect normal commit or rollback of an XA transaction because the user who started it knows its XID.
Argument suitability checking was improved for these spatial functions that test geometry relationships:
MBRWithin(). For more information, see Spatial Relation Functions That Use Object Shapes, and Spatial Relation Functions That Use Minimum Bounding Rectangles.
As a consequence of this work, performance of spatial relation functions was improved for cases where both arguments contain geometries of one dimensionality; that is, where the
GeometryCollectionarguments (or one
GeometryCollectionand another type of geometry) passed to a spatial relation function can be reduced to a
MultiPolygon. In those cases, the more complex general
GeometryCollectioncan be avoided. (Bug #22285402, Bug #79406)
--do-test-listoption for mysql-test-run.pl accepts an argument for a file containing tests one per line. Lines in the file are now accepted in any of the following formats:
test_name test_name.test suite_name.test_name suite_name.test_name.test path/to/test/file
mysql-test-run.pl and mysqltest are now more restrictive about permitted test case and result file names. Names must consist only of alphanumeric characters (
0-9), dash (
-), or underscore (
_), and cannot start with dash or underscore. An error occurs for nonconforming file names. (Bug #25487471)
These changes were made to the
--xml-reportoption for mysql-test-run.pl:
<failure>tag identifies any test failing on a retry attempt.
Separate statistics and fields are included for skipped and disabled tests.
The XML report is created in the build directory if no absolute path is given for out-of-source builds.
For combination runs, a field named
variationis included in the
Results for all tests belonging to a suite are aggregated within a single
Information regarding failures is reported with a brief reason in an attribute named
message, along with other details within the tag.
mysql-test-run.pl now supports a
--report-unstable-testsoption that has these effects:
Reports any test that has passed using at least one retry attempt in a separate category called “Unstable tests” in the summary.
If all failures are due to unstable tests, mysql-test-run.pl produces a warning but exits successfully.
Adds a new XML tag to report unstable tests, if the
--xml-reportoption is also specified.
(Bug #24473420, Bug #25984429)
mysqltest now supports a
replace_numeric_roundcommand that takes an argument value from 0 to 16 indicating the number of decimals to round numeric values to. This can be used to help prevent result content mismatch errors for tests in which slightly different results are obtained across platforms due to precision differences. Thanks to Daniel Black for the patch. (Bug #23280117, Bug #81399)
mysql-test-run.pl now looks for a
file, which is analogous to the
file but for specifying test-specific client options. (Bug #17084918)
X Plugin now handles expired SSL certificates correctly. (Bug #25835833)
The output of clauses such as
HAVINGhas been improved by moving the
grouping_criteriaexpression to the derived query. (Bug #25549637, Bug #24497007)
During install of the X Plugin a blank rule was being generated in the Firewall white list. (Bug #24488234)
Linux: The generic Linux build for MySQL 8.0 now supports Non-Uniform Memory Access (NUMA) on its host system. Systems that use the build need to have
libnumainstalled on them. See Installing MySQL on Unix/Linux Using Generic Binaries for more details. (Bug #26005558)
Incompatible Change; InnoDB:
REORGANIZE PARTITION, and
ALTER TABLEoptions are now supported by native partitioning in-place APIs and may be used with
EXCHANGE PARTITION, which previously supported
LOCKclauses, is performed by a new server layer API for compatibility with the MySQL data dictionary. Partition and table exchange is performed internally by the native partition handler.
ALGORITHM=INPLACEdeletes data stored in the partition and drops the partition. However,
old_alter_table=ONrebuilds the partitioned table and attempts to move data from the dropped partition to another partition with a compatible
PARTITION ... VALUESdefinition. Data that cannot be moved to another partition is deleted.
InnoDBnow uses tablespace map files during recovery to identify tablespaces that require redo log application. This is a change from redo log tablespace discovery that was used previously.
If tablespace map files are lost or corrupted, the
innodb_scan_directoriesstartup option may be used to specify tablespace directories when starting MySQL after a server outage.
For more information, see Tablespace Discovery During Crash Recovery. (Bug #24793413)
InnoDB: In previous releases,
KEY_BLOCK_SIZEattributes specified in
ALTER TABLEstatements even if those attributes are silently ignored when creating or altering a table. In such cases,
SHOW CREATE TABLEand the
Create_optionscolumn reported by
SHOW TABLE STATUSshow the specified
KEY_BLOCK_SIZEattributes rather than the actual attributes of the table.
InnoDBnow stores and reports the actual
KEY_BLOCK_SIZEattributes of the table.
Replication: In previous versions issuing
STOP GROUP_REPLICATIONstopped the plugin but the server still accepted transactions. This meant the transactions were not transmitted to the group. To make
STOP GROUP_REPLICATIONsafer, now
super_read_onlyis set to
ONimmediately upon issuing
STOP GROUP_REPLICATION, which ensures no transactions are accepted. (Bug #25495067, Bug #84795)
Replication: When there was a network partition and a member was in a minority all queries to that member blocked. To improve this situation, the
group_replication_unreachable_majority_timeoutvariable has been added which enables you to configure how long members in a minority wait to regain contact with a member in the majority before leaving the group. (Bug #25473794)
group_replication_transaction_size_limitvariable was added to enable you to protect a group against large transactions causing a failure. (Bug #84785, Bug #25510757)
Replication: Support for binary log files created by versions earlier than MySQL 5.0 has been removed, and binary log Version 1 and Version 3 formats are no longer supported by slaves or mysqlbinlog.
replication_group_member_statsPerformance Schema tables have been extended to provide more monitoring information for Group Replication. The
replication_group_memberstable now includes information about member's roles, and the versions of MySQL running on a member. The
replication_group_member_statstable now includes information about all members in the group, the applier and local queues, and rolled back transactions.
Replication: The receiver thread has been improved to no longer block other thread's activities when waiting for disk space. This improves the monitoring of replication ensuring that it reports correctly when the receiver thread is waiting for disk space. If you are not able to free disk space to allow the receiver thread to continue its activity, it can be forcefully stopped without side effects in most cases.
Replication: A new transaction length field has been added to the
Gtid_log_eventwhich stores the transaction length in bytes.
Replication: The following variables have had their defaults changed to ensure replication is as robust and efficient as possible by default:
transaction_write_set_extraction=XXHASH64enables the optimal method for generating write set hashes.
slave_rows_search_algorithms='INDEX_SCAN,HASH_SCAN'enables the use of a hash table to avoid repeated table scans when no Primary Key or Primary Key Equivalent (non-null unique secondary index) exists for a table
expire_logs_days=30causes unused binary logs that are older than 30 days to be purged, ensuring storage space is used efficiently.
Replication: The Group Replication plugin now notifies other components in the server that some relevant events have happened. Upon view changes, recovery state updates, network partitioning and primary election, the plugin informs listeners registered in the service registry and notifies them that an event has occurred. These listeners, which could even be other plugins, can then react to these events.
group_replication_member_weightvariable has been added which enables you to control the election of new primaries in single-primary mode. In previous versions primary election was based on the member's UUID, with the lowest UUID elected as the new primary in the event of fail over. Use this variable to assign numeric weights to members to ensure that specific members are elected, for example during scheduled maintenance of the primary or to ensure certain hardware is prioritized.
Replication: The following new system variables have been added to configure Group Replication:
These variables enable you to fine tune flow control of individual group members, adjusting the quota depending on the task the member performs in the group.
JSON: Added support for ranges in the XPath expressions used with many MySQL JSON functions, including
JSON_REMOVE(). Such a range is specified using the syntax
endare, respectively, the first and last indexes of a range of elements from a JSON array (always numbered starting with 0). For example,
$[1 to 3]includes the second, third, and fourth elements, as shown here:
mysql> SELECT JSON_EXTRACT('[1, 2, 3, 4, 5]', '$[1 to 3]'); +----------------------------------------------+ | JSON_EXTRACT('[1, 2, 3, 4, 5]', '$[1 to 3]') | +----------------------------------------------+ | [2, 3, 4] | +----------------------------------------------+ 1 row in set (0.00 sec)
This work also provides support in such expressions for the
lastkeyword, which you can use to represent the index of the last (rightmost) element in the current array, like this:
mysql> SELECT JSON_EXTRACT('[1, 2, 3, 4, 5]', '$[last]'); +--------------------------------------------+ | JSON_EXTRACT('[1, 2, 3, 4, 5]', '$[last]') | +--------------------------------------------+ | 5 | +--------------------------------------------+ 1 row in set (0.00 sec)
Indexes relative to the end of the array are also supported, as shown here:
mysql> SELECT JSON_EXTRACT('[1, 2, 3, 4, 5]', '$[last-4 to last-2]'); +--------------------------------------------------------+ | JSON_EXTRACT('[1, 2, 3, 4, 5]', '$[last-4 to last-2]') | +--------------------------------------------------------+ | [1, 2, 3] | +--------------------------------------------------------+ 1 row in set (0.00 sec)
For further information and examples, see Searching and Modifying JSON Values. (Bug #79052, Bug #22285926)
JSON: The optimizer now supports partial (in-place) updates of JSON documents when using
JSON_REMOVE(). (Previously, all updates of
JSONcolumn values were done by completely removing the previous document and writing the new one in its place.) In general, this optimization can be performed when the replacement value is less than or equal in size to the column's original value, and no new elements are added to the original value. It cannot be performed for a simple replacement of the column value of the form
The JSON Data Type provides more information about the requirements for an update to be optimized in this way.
This work includes the addition of two JSON utility functions,
JSON_STORAGE_FREE(), in the MySQL Server.
JSON_STORAGE_SIZE()returns the number of bytes used to store the binary representation of a JSON document, whether the document is presented as a column value in a table, as the value of a user variable, or as a JSON literal. In the case of a
JSONcolumn, this is the space used to store the JSON document as it was inserted into the column, prior to any partial updates that may have been performed on it since then.
JSON_STORAGE_FREE()shows the number of bytes in the binary representation of a JSON column value that were freed by the most recent partial update of the column. For a user variable storing a JSON document,
JSON_STORAGE_FREE()always returns 0; it also returns 0 if the argument is a JSON literal.
Each of these functions, like many other MySQL functions that act on JSON values, also accepts a string that can be successfully parsed as a JSON document. For more information and examples, see JSON Utility Functions.
.specfiles now include support for running unit tests. (Bug #25814143, Bug #85743)
The mysql client now supports a
--binary-as-hexoption that causes display of binary data using hexadecimal notation (
0x). Thanks to Daniël van Eeden for the patch. (Bug #25340722, Bug #84391)
SHOW TABLESstatement now supports an optional
EXTENDEDkeyword that causes statement output to list hidden tables created by failed
ALTER TABLEstatements. (These temporary tables have names beginning with
#sql.) For more information, see SHOW TABLES Statement. (Bug #24786075, Bug #83241)
Serialized dictionary information (SDI) files now have a lowercase extension (
.sdi) instead of uppercase.
TempTablestorage engine replaces the
MEMORYstorage engine as the default engine for in-memory internal temporary tables. The
TempTablestorage engine provides efficient storage for
internal_tmp_mem_storage_enginesession variable defines the storage engine for in-memory internal temporary tables. Permitted values are
TempTable(the default) and
temptable_max_ramconfiguration option defines the maximum amount of memory that can be occupied by the
TempTablestorage engine before it starts storing data on disk. The default value is 1073741824 bytes (1GiB).
memory/temptable/physical_diskPerformance Schema instruments may be used to monitor
TempTablememory allocation and disk storage.
The MySQL 8.0.2 release introduces a number of undo related changes:
The number of undo tablespaces can now be modified at runtime, or when the server is restarted, using the
innodb_undo_tablespacesconfiguration option. Previously, the number of undo tablespaces could only be configured when initializing the MySQL instance and could not be changed afterward. This change permits the addition of undo tablespaces and rollback segments as the database grows.
innodb_undo_log_truncateis enabled by default. When enabled, any undo tablespace that exceeds the threshold value defined by
innodb_max_undo_log_sizeis marked for truncation. See Truncating Undo Tablespaces.
innodb_undo_tablespacesdefault value was changed from 0 to 2, which means that rollback segments are created in two separate undo tablespaces instead of the
InnoDBsystem tablespace by default. A minimum of two undo tablespaces is required to permit truncation of undo logs.
innodb_undo_tablespacesto 0 is deprecated and will not be supported in a future MySQL version.
The naming convention used for undo tablespace files is changed from
is the undo space number.
innodb_rollback_segmentsconfiguration option defines the number of rollback segments per undo tablespace. Previously,
innodb_rollback_segmentswas a global setting that specified the total number of rollback segments for the MySQL instance. This change increases the number of rollback segments available for concurrent transactions. More rollback segments increases the likelihood that concurrent transactions use separate rollback segments for undo logs, resulting in less resource contention.
innodb_undo_logsconfiguration option is removed. The
innodb_rollback_segmentsconfiguration option performs the same function and should be used instead.
Innodb_available_undo_logsstatus variable is removed. The number of available rollback segments per tablespace may be retrieved using
SHOW VARIABLES LIKE 'innodb_rollback_segments';
SHOW INDEXstatements now support an optional
EXTENDEDkeyword that causes statement output to include information about hidden columns and indexes that MySQL uses internally and are not accessible by users. For more information, see SHOW COLUMNS Statement, and SHOW INDEX Statement.
Incompatible Change: Plugins such as Group Replication and X Plugin now use the
mysql.sessionaccount added in this version. If you are upgrading from a previous version which did not include the
mysql.sessionaccount you must run mysql_upgrade to ensure the account is created. If mysql_upgrade is not run, plugins fail to start with the error message There was an error when trying to access the server with user: mysql.session@localhost. Make sure the user is present in the server and that mysql_upgrade was run after a server update. Previously used users, such as
mysqlxsys, are no longer created. (Bug #26042764)
References: See also: Bug #24311527, Bug #25642343, Bug #25750822, Bug #25103980, Bug #83841.
InnoDB: Queries run on
INFORMATION_SCHEMA.TABLESperformed more slowly with
latest. (Bug #26197113)
ALTER TABLE ... ADD PARTITIONoperation that specified a
DATA DIRECTORYclause failed to ignore the
TABLESPACEattribute of the table. (Bug #26113652)
foreign_key_checksis disabled, a child table with a foreign key constraint can be created before the parent table, which can result in a foreign key constraint failure, as the parent table is unaware of the constraint. When a table is created, there is now a call to load foreign key constraints for the table and check for child tables. (Bug #25976199)
InnoDB: A parsing error occurred while optimizing a table with a full-text index. (Bug #25914332)
InnoDB: Compiling MySQL on Windows with Microsoft Visual C++ 2015 or macOS 10.12.4 with GCC 4.2.1 or Apple LLVM version 8.0.0 (clang-800.0.38) returned warnings. (Bug #25910531)
InnoDB: In debug builds, an assertion was raised during bootstrap when the system tablespace file (
ibdata1) ran out of space during creation of doublewrite pages. (Bug #25872368)
InnoDB: Incorrect locking order caused a deadlock when
InnoDBattempted to persist an auto-increment counter value to disk. (Bug #25833228)
InnoDB: Internal methods for accessing dictionary table object data did not account for virtual columns. (Bug #25822154)
References: This issue is a regression of: Bug #23748128.
InnoDB: The length of a virtual column field in a virtual index record was less than the expected template column length. (Bug #25793677)
InnoDB: In debug builds, shutting down the server with
--innodb-fast-shutdown=0raised an assertion. (Bug #25756224)
InnoDB: The ibd2sdi utility exited when run on an unsupported file type. (Bug #25738491)
InnoDBdid not set the compression algorithm when opening a partitioned table. (Bug #25685868)
InnoDB: A latch that was held while registering a file close caused a hang condition. (Bug #25658467)
InnoDB: During recovery, prepared transactions were rolled back if the
innodb_force_recoverysetting was greater than 0. (Bug #25651042)
InnoDB: Updates to data dictionary tables combined with updates to
InnoDBsystem tables for full-text search auxiliary tables raised a lock-related assertion. (Bug #25610353)
InnoDB: The server allocated memory unnecessarily for an operation that rebuilt the table. (Bug #25573565, Bug #85043)
InnoDB: Test-related code intended to simulate a random read on a nonexistent page raised an invalid assertion. (Bug #25479538)
References: This issue is a regression of: Bug #25053705.
InnoDBincorrectly printed messages to the error log when operations were performed on a table that referenced a nonexistent foreign key. (Bug #25365223)
InnoDB: During a
TRUNCATE TABLEoperation on a file-per-table tablespace, a dictionary operation lock was released before eviction of dirty pages from the buffer pool, causing a latch order violation. (Bug #25357789)
InnoDB: When using an index merge optimizer switch, a
SELECT COUNT(*)operation sometimes returned 0. Partitioning code incorrectly performed a
memcpyinstead of a column copy of columns read by the index, causing the wrong records to be copied. (Bug #25332330, Bug #81031)
InnoDB: After a
RENAME TABLEoperation that moved a table to a different schema,
InnoDBreturned an error on restart indicating that it could not locate the tablespace data file.
InnoDBfailed to update
INNODB_SYS_DATAFILESdata dictionary table during the
RENAME TABLEoperation. (Bug #25189192, Bug #84038)
InnoDB: During an
ALTER TABLEoperation that rebuilt a table containing a virtual column,
InnoDBfailed to apply a concurrent insert log record. (Bug #24961167)
InnoDBfailed to apply the concurrent delete log for an in-place
ALTER TABLEoperation due to a virtual column validation issue. (Bug #24960450)
Partitioning: Following execution of an
ALTER TABLE ... OPTIMIZE PARTITIONstatement, names of tablespaces used by the table were not preserved, and subsequent accesses to the table eventually caused the server to fail. (Bug #25512556)
Replication: When replicating a partitioned table with an index, on a replication slave where
HASH_SCANwas specified as part of the
slave_rows_search_algorithmssetting, the slave I/O thread sometimes stopped with an error
HA_ERR_KEY_NOT_FOUND. (Bug #26137159)
USEstatement that followed a
SET GTID_NEXTstatement sometimes had no effect. (Bug #26128931)
Replication: A misleading warning was issued when the command FLUSH LOGS or PURGE LOGS BEFORE was used on a binary log file with an expiry time set, and the binary log file was in use. The warning related to the file being in use, and implied that a purge attempt had taken place, even if the expiry time had not yet been reached. Now, MySQL checks the expiry time of the binary log file first, and only then checks whether the file is in use. So the warning is only issued for an in-use binary log file that is old enough to be purged. (Bug #25973525)
Replication: Groups can now contain members running different server versions to enable you to do online upgrades of a replication group. The rules for combining members in a group with different versions are:
If you have a group with 8.0 members, you cannot add a 5.7 member
If you have a group with 5.7 members you can add a 8.0 member, but it remains in read-only mode. Writing to this member is dangerous while the group contains multiple server versions and should be avoided.
In a single-primary group, if the current primary leaves the group and a new primary must be elected, the primary is first chosen from the lower version members. If no lower version member is found, the primary is chosen from newer version members. (Bug #25876807)
binlog_checksum=NONEwas set on a MySQL server after startup, and then Group Replication was started, if an error occurred, the server remained in
RECOVERINGstate and could not be shut down. (Bug #25793366, Bug #85667)
Replication: Now that XA transactions are prepared and committed in two parts, an issue with statement-based replication has been identified. If two XA transactions committed on the master are being prepared on the slave in the inverse order, locking dependencies can occur that cannot be safely resolved. The issue is not present with row-based replication.
XA transactions are therefore now considered unsafe for statement-based replication.
binlog_format = STATEMENT, a warning is issued for DML statements inside XA transactions, and replication might fail with deadlock on slaves.
binlog_format = MIXED, DML statements inside XA transactions are logged using row-based replication.
binlog_format = ROW, DML statements inside XA transactions are logged as before.
(Bug #25786490, Bug #85639)
Replication: The following Performance Schema replication tables now continue to be populated when the Performance Schema is disabled:
The exception is local timing information (start and end timestamps for transactions) in the replication tables
replication_applier_status_by_worker. This information is not collected when the Performance Schema is disabled. (Bug #25694813)
Replication: In a Group Replication setup where circular asynchronous replication was implemented between members of different replication groups, view change log events were repeatedly replicated between the groups with new generated GTIDs each time. The fix ensures that view change log events are ignored outside the named replication group where they occur, and never generate new GTIDs. (Bug #25674926)
References: See also: Bug #26049695, Bug #25928854, Bug #25721175.
Replication: When first starting the MySQL server following an installation from RPM, passwword validation plugin is activated by default (true only for RPM installations). If binary logging was already enabled at this time, the activation was logged, even though plugin activations should not be recorded in the binary log. (Bug #25672750)
XA COMMIT, and
XA ROLLBACKstatements, used to perform XA transactions, were incrementing the counter for a
replicate-do-dbfilter that named the relevant database. The counter is no longer incremented for these statements. (Bug #25669344)
Replication: In a setup where single-primary Group Replication was combined with asynchronous replication, for example with S1 and S2 forming a group and with S2 and S3 functioning as master and slave, secondaries such as S2 were accepting transactions and these could then enter the group. The fix prevents secondaries creating an asynchronous replication channel when belonging to a single-primary group, and Group Replication cannot be started when asynchronous replication is running. (Bug #25574200, Bug #85047)
References: See also: Bug #86325, Bug #26078602.
Replication: Group Replication failed to start if the
super_read_onlysystem variable was enabled. (Bug #25481287, Bug #84733)
Replication: MySQL 8.0 servers were not able to join a replication group that contained MySQL 5.7 servers, and MySQL 5.7 servers in a replication group could not be upgraded to MySQL 8.0. (Bug #25477979)
Replication: In the event that a member failed to join a group the member was not stopping and continued to accept transactions. To avoid this set your members to have
my.cfgfile. Group Replication now checks for this setting upon successful start up and sets
super_read_only=0. This ensures that members which do not successfully join a group cannot accept transactions. (Bug #25474736, Bug #84728)
Replication: When mysqlbinlog output from a MySQL 5.7 server was applied on a MySQL 8.0 server, the MySQL 8.0 server generated its own timestamp for the
original_commit_timestamp. The fix ensures that the value of
original_commit_timestampis correctly set to 0, meaning that the timestamp is not known. (Bug #25316086)
Replication: If the server was started with the
--log-binoption, discovery of one Performance Schema table with invalid structure caused all subsequently checked Performance Schema tables to be marked invalid as well. (Bug #25041396)
Replication: Interleaved transactions could sometimes deadlock the slave applier when the transaction isolation level was set to
REPEATABLE-READ. (Bug #25040331)
slave_skip_errorssystem variable did not permit error numbers larger than 3000. Thanks to Tsubasa Tanaka for the patch. (Bug #24748639, Bug #83184)
Replication: The SQL state of
HY00, which is generic and not suitable for rollback. The SQL state has been modified to 40000. (Bug #24658431)
Replication: mysqlbinlog, if invoked with the
--rawoption, does not flush the output file until the process terminates. But if also invoked with the
--stop-neveroption, the process never terminates, thus nothing is ever written to the output file. Now the output is flushed after each event. (Bug #24609402)
Replication: A memory leak in mysqlbinlog was fixed. The leak happened when processing fake rotate events, or when using
--rawand the destination log file could not be created. The leak only occurred when processing events from a remote server. Thanks to Laurynas Biveinis for his contribution to fixing this bug. (Bug #24323288, Bug #82283)
Replication: Loading and initialization of the Group Replication plugin failed if no
rootaccount was present. (Bug #24311527)
References: See also: Bug #25750822, Bug #25103980, Bug #83841.
Replication: A slave server could lose events not yet applied when
MASTER_AUTO_POSITION=0, both replication threads were stopped, and the applier delay was changed using
CHANGE MASTER TO MASTER_DELAY=. (Bug #23203678, Bug #81232)
References: See also: Bug #25340185, Bug #84375.
Replication: Transmission of large GCS messages could take so long the sender appeared to have died. (Bug #22671846)
Replication: Multithreaded slaves could not be configured with small queue sizes using
slave_pending_jobs_size_maxif they ever needed to process transactions larger than that size. Any packet larger than
slave_pending_jobs_size_maxwas rejected with the error ER_MTS_EVENT_BIGGER_PENDING_JOBS_SIZE_MAX, even if the packet was smaller than the limit set by
With this fix,
slave_pending_jobs_size_maxbecomes a soft limit rather than a hard limit. If the size of a packet exceeds
slave_pending_jobs_size_maxbut is less than
slave_max_allowed_packet, the transaction is held until all the slave workers have empty queues, and then processed. All subsequent transactions are held until the large transaction has been completed. The queue size for slave workers can therefore be limited while still allowing occasional larger transactions. (Bug #21280753, Bug #77406)
Replication: An incident event that broke replication was not written to the binary log with a GTID, so that it was not possible to skip the event using
SET gtid_next=. Instead, it was necessary to set the relay log file and relay log positions directly; this meant that, when autopositioning was enabled, it was necessary first to disable it, then to set the relay log file and position, and finally to re-enable autopositioning.
Now in such cases MySQL writes the incident event into the statement cache, so that a GTID is generated and written for it prior to flushing, and that the slave applier works with the change. Then users can skip the event using the SQL statement
SET gtid_next=, followed by
COMMIT. (Bug #19594845)
SHOW SLAVE STATUS FOR CHANNEL 'group_replication_recovery'following a restart of a server using group replication led to an unplanned shutdown. (Bug #85739, Bug #25813258)
Replication: Setting an empty filter rule using
CHANGE REPLICATION FILTERcaused issues when running tests with UBSan. (Bug #85405, Bug #25702297)
Replication: Indexes for the Performance Schema
replication_applier_status_by_workertable worked correctly only for single threaded replication. This was due to the fact that the indexing relied exclusively on the SQL thread ID regardless of the existence of any workers. Now in such cases, the index also considers worker thread ID when multithreading is enabled on the slave. (Bug #84646, Bug #25444610)
Replication: In certain cases, the master could write to the binary log a
last_committedvalue which was smaller than it should have been. This could cause the slave to execute in parallel transactions which should not have been, leading to inconsistencies or other errors. (Bug #84471, Bug #25379659)
Replication: When using
group_replication_ip_whitelist=AUTOMATIC, IPs in the private network are permitted automatically, but some class C IP addresses were not being permitted correctly. (Bug #84329, Bug #25503458)
Replication: When an existing GTID_NEXT transaction was assigned a conflicting GTID by the server, Group Replication generated an assert upon detecting two transactions with same GTID. This was because Group Replication generates the GTID after conflict detection, which is later than with master/slave replication. The fix relaxes some conditions to only be called when commit is done and a message has been added to alert you when a GTID has already been used. (Bug #84153, Bug #25232042)
Replication: The replication applier thread returns Error 3002
ER_INCONSISTENT_ERRORwhen there is a difference between an expected error number and the actual error number. It is now possible to ignore this error by using 3002 with
slave_skip_errors. (Bug #83186, Bug #24753281)
Replication: MySQL lost its GTID position following a restart when a dump from mysqldump had been used to load data.
To keep this problem from occurring, the
mysql.gtid_executedtable is now excluded automatically from dumps made by
mysqldump. (Bug #82848, Bug #24590891)
References: See also: Bug #87455, Bug #26643180.
Replication: It was possible to set
binlog_formatwithout causing an error when there were open temporary tables. (Bug #82467, Bug #24411680)
Replication: Corruption of relay logs for one channel in multi-source replication caused good channels not to be initalized during a server restart. In addition, when run with
--skip-slave-start=false, the server also failed to start slave threads for those channels which were in good condition, despite the fact that it should have started the slave threads for all good channels.
Now, regardless of any errors on other channels, the server attempts to create and initialize channels that are in good condition, and starts slave threads for the good channels if
--skip-slave-startis disabled. As part of this fix,
STOP SLAVE, which are intended to operate on all channels, are also modified such that they continue executing on all good channels even if they find bad channels among them. (Bug #82209, Bug #24285104)
Replication: The SQL thread was unable to GTID skip a partial transaction. (Bug #81119, Bug #25800025)
Replication: It was possible for
FLUSH LOGSto write a
Rotate_log_eventinto an uninitialized log file. (Bug #80368, Bug #22732184)
References: See also: Bug #23531998, Bug #81734.
Microsoft Windows: On Windows, the
Docs/INFO_SRCfile was missing the
build-dateentry. (Bug #25799855)
NULLIF()failed with an assertion error in debug builds if the result from this function was used in a JSON context. This result can now be used safely as a JSON value.
This fix includes using
DBUG_ASSERT()for handling this error instead of
DBUG_ABORT(), which caused debugging problems on some platforms. (Bug #25818544)
References: See also: Bug #21383497, Bug #21383530.
JSON: The internal
Json_arraynow uses a
std::vectorinstead of a
Prealloced_arrayto store its elements, which reduces the amount of heap space required by them. (Bug #85877, Bug #25867454)
When the character set of one string comparison operand was a superset of the character set of the other operand, some comparisons were disallowed that should be permitted by converting the operand with the “smaller” character set to the “larger” character set.
utf32are considered to be a superset of any other encoding. (Bug #27897053, Bug #25642319, Bug #85224)
mysqld_pre_systemdscript in RPM packages found the error log setting in option files if specified as
log-errorbut not as
log_error, though both are permitted. (Bug #26148391, Bug #86466)
ALTER TABLE ... ADD PARTITIONstatements could cause a server exit. (Bug #26132947)
SET PERSISTdid not work for X Plugin system variables. (Bug #26115672)
REFERENCESprivilege checking could use the incorrect database in some cases. (Bug #26106655)
gtid_modesystem variable not set in a
.cnfoption file or on the command-line, the server failed to restart after using
SET GLOBALto set
ON_PERMISSIVE. (Bug #26085712)
During prepared statement execution, too many bytes of a buffer could be read. (Bug #26042934, Bug #85937)
Dynamic privileges were not loaded when the server was started using the
--initializeoption. (Bug #26005645)
Debian client packages were missing information about conflicts with akonadi-backend-mysql packages. (Bug #26002288)
mysqldump could write database names in
USEstatements incorrectly. (Bug #25998635)
mysql_stmt_close()C API function was called, it freed memory that later could be accessed if
mysql_stmt_sqlstate()was called. To obtain error information after a call to
mysql_sqlstate()instead. (Bug #25988681)
Upgrading from MySQL 5.7 to MySQL 8.0 failed during data dictionary creation due to table names that exceeded the 64 character length limit. (Bug #25973237, Bug #86120)
Queries could be cached incorrectly, leading to incorrect query results, under these circumstances:
InnoDBtable; rows are being inserted but have not yet been committed; a query uses the table as a base table in a derived table; the optimizer chooses to materialize the derived table. (Bug #25943038, Bug #86047)
mysqld misbehaved if compiled with AddressSanitizer enabled and invoked with
--basedir=(that is, with an empty option value). (Bug #25914296, Bug #85994)
Attempting to drop a data file from a tablespace with multiple data files returned a duplicate file name error. (Bug #25858461)
An attribute was added to data dictionary client method declarations to generate compilation warnings when return values are ignored. (Bug #25840927)
The data dictionary class hierarchy was simplified. (Bug #25835968, Bug #85811)
A restriction that prevented the data dictionary object update function (
Dictionary_client::update) from being called twice on the same object was removed. (Bug #25833932, Bug #85800)
On Ubuntu platforms, the MySQL service script did not terminate correctly if the user running the script did not have permission to access the data directory or PID file. (Bug #25825833)
String comparison queries on the
INFORMATION_SCHEMA.CHARACTER_SETS.DESCRIPTIONfield were case sensitive, which is a regression from earlier releases. (Bug #25824297)
Man pages for a few utilities were missing from Debian/Ubuntu packages. (Bug #25811814)
INFORMATION_SCHEMAview executed in
LOCK TABLESmode, using the
SERIALIZABLEisolation level, and with
autocommitdisabled, failed to use non-locking reads, which could block DDL operations or cause deadlocks. (Bug #25811413)
Table locking failures could occur if tables were used by a trigger for which a trigger of the same name existed in another database and the database names differed only in lettercase. (Bug #25807393)
An in-place MySQL upgrade failed if a
.TRGfile was missing the
createdline for the trigger definition. (This problem is unlikely to occur for upgrades to MySQL 8.0 from 5.7 because the
createdline is present for MySQL 5.7.2 and higher.) (Bug #25805260, Bug #85704)
field-tunit test failed to run with AddressSanitizer enabled. Thanks to Laurynas Biveinis for the patch. (Bug #25803823, Bug #85678)
Debian client packages were missing information about conflicts with native packages. (Bug #25799475)
After importing a table with
DECIMALcolumn, accessing the table raised an assertion. (Bug #25792649)
ALTER TABLEcould fail when the default character set changed to
utf8mb4due to incorrect column length calculations. (Bug #25779239, Bug #85614)
If a server component
deinit()method failed, the component was still unloaded. (Bug #25764325)
Stored program execution could fail for
DEFINERaccounts with expired passwords even if they were locked and could not be used to connect to the server. Now
DEFINERaccounts are not checked for expired passwords if they are locked. (Bug #25741966)
DISTINCToperations on temporary tables could produce incorrect results due to allocation of too-small comparison keys. (Bug #25740550, Bug #85518)
In optimizer trace output,
num_tmp_filesdid not actually indicate number of files. It has been renamed to
num_initial_chunks_spilled_to_diskand indicates the number of chunks before any merging has occurred. (Bug #25733784, Bug #85487)
The Perl path in
#!lines at the beginning of Perl scripts has been adjusted to
/usr/local/bin/perlfor FreeBSD 11. (Bug #25719975)
IGNORE_SPACESQL mode enabled, syntax error messages always reported line 1 as the line number. (Bug #25717617)
mysqldump failed to properly quote certain identifiers in SQL statements written to the dump output. (Bug #25717383)
Dropping a tablespace without that associated tablespace file failed with an error stating that the tablespace dictionary object is invalid. (Bug #25717019)
Client preauthorization by the server was missing a length check for a length-encoded string. (Bug #25714674)
In debug builds, a
CREATE TABLESPACEoperation raised an invalid assertion when using the NDB storage engine. A validation function that checked for zero-length data files did not apply to NDB tablepaces and was removed. (Bug #25700242)
For debug builds,
EXPORT_SET()operations could raise an assertion for some arguments. (Bug #25688192)
For debug builds, the assertion added for Bug#59686 was too strict and could be raised when it should not have been. (Bug #25685958)
References: See also: Bug #59686.
A code refactoring in MySQL 8.0.1 caused several assertions to be raised in debug builds. (Bug #25669590, Bug #25669606, Bug #25669580, Bug #25688504)
References: This issue is a regression of: Bug #25221172, Bug #84103.
For some double-precision calculations, overflow could occur when calculating the exponent part. (Bug #25664323, Bug #85290)
SHOW INDEXoutput was inconsistent for a
FULLTEXTindex defined on multiple columns. (Bug #25659276)
During a group commit, the stack could incorrectly be reported exhausted, leading to stack overflow. (Bug #25656875)
An in-place (binary) upgrade could change the default
character_set_connectionvariables for stored routines, possibly resulting in different routine behavior. (Bug #25633041)
For debug builds, the partitioning handler could inappropriately evaluate generated column expressions in
UPDATEstatements, resulting in a raised assertion. (Bug #25615803, Bug #85179)
In strict SQL mode, an
ER_TRUNCATED_WRONG_VALUEerror could be converted from a warning to an error but then ignored, leading to a raised assertion. (Bug #25586959, Bug #25586673)
Failure occurred for
ALTER TABLEon an
ARCHIVEtable containing a
NOT NULLcolumn having a geometry data type. For debug builds, an assertion was raised. For non-debug builds, an error occurred. (Bug #25582178, Bug #85059)
For debug builds, an assertion could be raised for
DROP TRIGGERof a trigger for a table used by a view. (Bug #25581925)
For upgrades from MySQL 5.7 to MySQL 8.0, the server wrote unnecessary parsing warnings to the error log. These messages are now suppressed. (Bug #25518436, Bug #84889)
The server exited abnormally attempting to access invalid memory. (Bug #25501659)
The maximum length of the
namefield in the
column_type_elementstable in the data dictionary was extended from 255 bytes to 1020 bytes to accommodate long
SETvalues that could result from using a multibyte character set.
When using a multibyte character set, a single
SETelement occupies a maximum of
wbytes in the column type definition in the data dictionary, where
Mis the element literal length and
wis the number of bytes required for the maximum-length character in the character set.
In prior releases, the maximum supported length of an individual
SETelement depended on the number of elements in the type. Thus, there could be a single element with (
w) = 64K, or 64K elements with (
w) = 1.
The maximum supported length of an individual
SETelement is now
M<= 255 and (
w) <= 1020, regardless of the number of elements in the type. (Bug #25481355)
mysqldsecured by TCP wrappers and the
hosts.denyfiles configured to restrict access from an IP address, connection attempts from that address resulted in too many messages to the error log. (Bug #25476479, Bug #84708)
mysqlpumpno longer includes the
slave_relay_log_infotables in dumps of the
mysqlsystem database. Restoring a dump file containing these tables caused problems by changing the replication state improperly. (Bug #25469190)
Changes made by calling
MYSQL_OPT_SSL_MODEcould be affected by later
mysql_options()calls. Now setting
MYSQL_OPT_SSL_MODEis unaffected by later
mysql_options()calls. (Bug #25452210)
A race condition could occur for
CREATE TABLEstatements with
INDEX DIRECTORYclauses. (Bug #25451091)
With AddressSanitizer enabled, compiling the
keyring_fileplugin produced One-Definition Rule violations. (Bug #25448205)
ALTER TABLE ... MODIFYon a
DATETIME NOT NULLcolumn using an
AFTERclause resulted in an
ER_INVALID_USE_OF_NULLerror. (Bug #25385334)
The range optimizer could create an incorrect query tree, resulting in a server exit. (Bug #25369742, Bug #25586531)
mysqld_failed to start the server if the
--datadiroption was specified with a relative path name. (Bug #25364806)
XA ROLLBACK, and
XA COMMITfor a transaction from a disconnected session did not take a global commit lock and modified the binary log and
InnoDBredo log even when
FLUSH TABLES WITH READ LOCKwas in effect. This could lead to inconsistent backups when backup tools assumed that the server was in a read-only state. (Bug #25364178, Bug #84442)
The fix for Bug #78777 had different effects depending on whether the Performance Schema is enabled. (Bug #25309017, Bug #84305)
References: This issue is a regression of: Bug #78777.
An aggregate function in some nested queries could cause a server exit. (Bug #25303711)
Virtual generated column expressions that used the
HEX()functions could cause problems if the connection character set was changed. In this context, the table character set is now used for these functions regardless of connection character set. (Bug #25287633)
Rewriterplugin did not perform locking properly if the
read_onlysystem variable was enabled. (Bug #25264253)
For a table having a
DATETIMEcolumn having a default of
CURRENT_TIMESTAMP, the column could be initialized to
0000-00-00 00:00:00'if the table had a
BEFORE INSERTtrigger. (Bug #25209512, Bug #84077)
SHOW PROCESSLISToutput drifted higher over time. (Bug #25101724, Bug #83019)
An assertion was raised when a create function at server bootstrap attempted to evaluate non-resolved expressions. (Bug #24961932)
The number of data dictionary cache lookups was reduced, and
ALTER TABLEhandling of triggers was improved help ensure trigger persistence in case of a server exit during
ALTER TABLEprocessing. (Bug #24930129, Bug #83473)
Certain queries against
InnoDBtables that used a primary key and a subquery could return incorrect results if the
index_merge_intersectionflag of the
optimizer_switchsystem variable was enabled. (Bug #24829050, Bug #79675)
On x86 machines, the
uint3korr()macro read 4 bytes of data instead of the intended 3 bytes. (Bug #24807826, Bug #83264)
An assertion was raised during a fetch operation by the memcached plugin. (Bug #24605783)
Queries that contained
UNIONin a subquery and
GROUP BYcould return incorrect results. (Bug #24595639)
Some syntactically incorrect
CREATE INDEXstatements could cause a server exit rather than a syntax error. (Bug #24593992)
An in-place MySQL upgrade failed if the
innodb_index_statstables were not present in the
mysqlsystem database. (This problem should not occur for upgrades to MySQL 8.0 from 5.7 because those tables are present in MySQL 5.7.) (Bug #24557143)
Incorrect behavior could occur for
INSERTstatements executed in stored-program or prepared-statement context, if the
VALUESpart of an
ON DUPLICATE KEY UPDATEclause referred to a
BLOBvalue in the
INSERTcolumn list. (Bug #24538207, Bug #25361251, Bug #25530880, Bug #25684790)
systemd support scripts in Debian packages contained hardcoded references to the data directory, making it difficult to change the data directory using
--datadir. (Bug #24398446, Bug #82417)
MySQL failed to compile under macOS 10.10.5 using Clang. (Bug #24352163, Bug #82340)
REPLACEstatement tried to update a row in a table containing a virtual generated column of type
BLOB, subsequent DML statements could behave incorrectly. (Bug #23573575)
PROXYgrants were not replicated to slaves, causing incorrect replication. (Bug #23289541, Bug #81424, Bug #23623115)
The help output from mysqlxtest has been improved. (Bug #23107137, Bug #81086)
For builds with AddressSanitizer enabled, the
ST_Simplify()function could attempt to use already freed memory. (Bug #23023817)
Compiler flags were adjusted to eliminate numerous warnings that occurred when compiling the
keyring_fileplugin using Clang. (Bug #22834591, Bug #80524)
If enabling the Event Scheduler caused an event defined as
ON COMPLETION NOT PRESERVEto be dropped because its execution time had passed, the drop event was not written to the binary log, causing slaves not to replicate it and replication failure if an event of the same name was created later. (Bug #22150112)
CREATE TABLE ... SELECTstatements, adding an
ORDER BYclause changed column data types, or caused an assertion to be raised for debug builds. (Bug #16833464)
LOAD XMLperformance became noticeably slower when the XML file being read contained a great many spaces, such as those introduced by indenting or pretty-printing. Now all leading whitespace is trimmed from each such value before reading it into memory. (Bug #16212207)