MySQL 5.7.13

What's new in this version:

ACCOUNT MANAGEMENT NOTES:
- In MySQL 5.7.8, the maximum length of MySQL user names was increased from 16 to 32 characters, but some applicable contexts for this increase were overlooked. Additional changes in maximum user name length now have been applied.
- If you upgrade to this MySQL release from an earlier version, you must run mysql_upgrade (and restart the server) to incorporate the changes to the mysql system database. A server from MySQL 5.7.13 or higher for which mysql_upgrade has not been run continues to permit a maximum of 77 characters in the system tables mentioned previously, and an ER_USER_COLUMN_OLD_LENGTH error will occur in those system table contexts where a user_name@host_name value from 78 to 93 characters long is given.

AUDIT LOG NOTES:
- The audit_log plugin that forms the basis for MySQL Enterprise Audit now supports fine-grained filtering of events. This enables a reduction in the number of log entries written to the audit log file, which increases overall performance due to fewer write operations during log recording. It also simplifies subsequent audit log processing in terms of readability and processing time.
- Fine-grained audit log filtering is rule based, implemented using tables that store filter definitions and a set of user-defined functions (UDFs) that enable filter manipulation. To simplify installing the tables and UDFs along with the audit_log plugin, an installation script is now provided. For more information, see Installing or Uninstalling MySQL Enterprise Audit, and Audit Log Filtering.
- By default, audit log filtering now logs no auditable events for any users. This differs from the pre-MySQL 5.7.13 legacy audit log, which logs all auditable events for all users. To produce log-everything behavior, see the installation instructions.

CONFIGURATION NOTES:
- On platforms for which systemd support is installed, systemd has the capability of managing multiple MySQL instances. For details, see Configuring Multiple MySQL Instances Using systemd. Consequently, mysqld_multi and mysqld_multi.server are not installed because they are unnecessary. (Bug #81093, Bug #23134620)

SECURITY NOTES:
- The linked OpenSSL library for the MySQL Commercial Server has been updated to version 1.0.1t. Issues fixed in the new 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 #23229564)
- MySQL Server now includes an SQL interface for keyring key management, implemented as a set of user-defined functions (UDFs) that access the functions provided by the internal keyring service. For more information, see Keyring Key Management Functions. For information about the keyring service functions invoked by the UDFs, see The Keyring Service. For general keyring information, see The MySQL Keyring.

FUNCTIONALITY ADDED OR CHANGED:
- For better separation of output from multiple statements, mysqltest now flushes output sent to stdout when the output destination is not a file. Previously, flushing occurred only for file output. (Bug #21435906)
- support-files/MacOSX/ReadMe.txt is no longer included in MySQL distributions. (Bug #81038, Bug #23088916)
- The version of the tcmalloc library included in MySQL distributions was very old. It has been removed and is no longer included with MySQL. (Bug #80994, Bug #23068660)
- The my_make_scrambled_password() function in the C client library was restricted earlier in MySQL 5.7 (not exported to client programs). The function has once again been made visible to client programs. (Bug #80974, Bug #23061746)
- It is possible to use ALTER TABLE to change the default value of a column col_name, which may change the value of a generated column expression that refers to the column using DEFAULT(col_name). For this reason, ALTER TABLE operations that change the definition of a column now cause a table rebuild if any generated column expression uses DEFAULT(). (Bug #80299, Bug #22680839)
- This release adds an unquoting extraction operator ->>, sometimes also referred to as an inline path operator, for use with JSON documents stored in MySQL. The new operator is similar to the -> operator, but performs JSON unquoting of the value as well.
- The ->> operator can be used in SQL statements wherever JSON_UNQUOTE(JSON_EXTRACT()) would be allowed. This includes (but is not limited to) SELECT lists, WHERE and HAVING clauses, and ORDER BY and GROUP BY clauses.
- For more information, see Functions That Search JSON Values, and JSON Path Syntax. (Bug #78736, Bug #21980346)

BUGS FIXED:
- InnoDB: A rollback operation run concurrently with an operation involving a generated virtual column caused a server exit. (Bug #23313102)
- References: This issue is a regression of: Bug #21869656.
- InnoDB: Potential buffer overflow issues were corrected for the InnoDB memcached plugin. (Bug #23187607)
- InnoDB: An ALTER TABLE operation that added an AUTO_INCREMENT column on a table with virtual columns raised an assertion. (Bug #23052231)
- InnoDB: Statements executed in a transaction that was rolled back asynchronously by a higher priority transaction caused a deadlock error and subsequent replication failure. (Bug #23021168, Bug #80898)
- InnoDB: An ALTER TABLE operation that attempted to add a generated virtual column and a full-text index raised an assertion. (Bug #23014521)
- InnoDB: The full-text index cache was freed during a background index cache synchronization. (Bug #22996488)
- InnoDB: A full-text index operation raised an assertion. (Bug #22963169)
- InnoDB: Memory was allocated to I/O slots unnecessarily, causing an apparent memory leak. (Bug #22956469, Bug #80772)
- InnoDB: An ALTER TABLE operation raised an assertion when attempting to create a key containing a generated column. (Bug #22951879)
- InnoDB: A startup failure due to an invalid option resulted in a server exit after a subsequent restart. An error returned by the innobase_start_or_create_for_mysql function was not checked. (Bug #22939581, Bug #80761)
- InnoDB: An ALTER TABLE ... IMPORT TABLESPACE operation on file-per-table tablespace containing an encrypted table failed when run in a different session than the preceding ALTER TABLE ... DISCARD TABLESPACE operation. (Bug #22918999, Bug #80708)
- InnoDB: A FLUSH TABLES operation on a table with a discarded tablespace raised an assertion. (Bug #22899690, Bug #80669)
- InnoDB: A DML operation involving a table with a virtual column raised an assertion. (Bug #22899305)
- InnoDB: An invalid read at innobase_get_computed_value() raised a Valgrind error. (Bug #22898168, Bug #80667)
- InnoDB: Online DDL operations like ALTER TABLE ... ADD INDEX were not permitted for tables created with the ENCRYPTION attribute. (Bug #22897921)
- InnoDB: InnoDB memcached code assumed the nonexistence of the htonll() function on OS X, resulting in a build failure on OS X 10.10 and later. (Bug #22865112)
- InnoDB: Starting the server in read-only mode failed when encrypted tables were present. The call to fil_encryption_rotate() was not skipped when the server was started in read-only mode. (Bug #22723797)
- InnoDB: An R-tree purge operation raised an assertion. (Bug #22698076, Bug #80327)
- InnoDB: An INSERT operation on a table with a FULLTEXT index and FTS_DOC_ID column failed because the inserted FTS_DOC_ID value exceeded the permitted gap between consecutive FTS_DOC_ID values. To avoid this problem, the permitted gap between the largest used FTS_DOC_ID value and new FTS_DOC_ID value was raised from 10000 to 65535. (Bug #22679185)
- InnoDB: Validation code for transparent page compression incorrectly permitted innodb_strict_mode=OFF, which allowed the COMPRESSION attribute to be applied to a general tablespace. Page compression is only supported with file-per-table tablespaces. (Bug #22615096, Bug #80182)
- InnoDB: An memory order issue related to atomic operations caused assertion failures on ARM64 and POWER platforms. (Bug #22608616)
- InnoDB: DROP TABLESPACE returned an error if the remote general tablespace data file was missing. (Bug #22232892, Bug #79330)
- InnoDB: An ALTER TABLE operation that changed table row format from COMPACT to COMPRESSED raised an assertion. A function involved in the operation passed incorrect page size information. (Bug #22046353)
- InnoDB: With innodb_autoinc_lock_mode=0, multiple threads waiting for a table-level lock caused an unexpected deadlock. (Bug #21983865, Bug #78761)
- InnoDB: A race condition in trx_kill_blocking() raised an assertion. (Bug #21508537)
- InnoDB: An OPTIMIZE TABLE operation on a table with a full-text index raised an assertion. (Bug #21378944)
- InnoDB: A buffer pool load operation resulted in a “Cannot allocate 0 bytes” error. (Bug #21371070)
- InnoDB: A FLUSH TABLES ... FOR EXPORT operation appeared to stall. A loop in the ibuf_contract_in_background function failed to exit. (Bug #21133329, Bug #77011)
- InnoDB: A full-text query raised an assertion. Under certain circumstances, DDL operations such as ALTER TABLE ... RENAME caused full-text auxiliary tables to be removed on server restart. (Bug #13651665)
- Replication: With gtid_mode=ON, executing an empty query before setting gtid_next made the latter action fail. It was because only statements that do not change the data can be executed before one can successfully change gtid_next, and an empty query was not considered one of those “safe” statements. This fix allows the setting of gtid_next after an empty query. (Bug #22811150)
- Replication: Slaves running MySQL 5.7 could not connect to a MySQL 5.5 master due to an error retrieving the server_uuid, which is not part of MySQL 5.5. This was caused by changes in the method of retrieving the server_uuid. (Bug #22748612)
- References: This issue is a regression of: Bug #21455603.
- Replication: Setting certain semisynchronous-replication configurations on the master server when semisynchronous replication was not enabled on it might cause the server to exit. This fix prevents the unexpected exits to occur in the situation. (Bug #22602324)
- Replication: The variable explicit_defaults_ts was not initialized during the construction of a Query_event object. That caused Valgrind warnings for dependency on an uninitialized variable. This fix makes sure the variable is initialized. (Bug #22110916, Bug #78999)
- References: This issue is a regression of: Bug #18885916, Bug #72794.
- Replication: In the next_event() function, which is called by a slave's SQL thread to read the next even from the relay log, the SQL thread did not release the relaylog.log_lock it acquired when it ran into an error (for example, due to a closed relay log), causing all other threads waiting to acquire a lock on the relay log to hang. With this fix, the lock is released before the SQL thread leaves the function under the situation. (Bug #21697821)
- References: See also: Bug #20492319.
- Replication: A partially failed statement was not correctly consuming an auto-generated or specified GTID when binary logging was disabled. The fix ensures that a partially failed DROP TABLE, a partially failed DROP USER or a partially failed DROP VIEW consume respectively the relevant GTID and save it into @@GLOBAL.GTID_EXECUTED and mysql.gtid_executed table when binary logging is disabled. (Bug #21686749)
- Replication: An intermittent ASan error was being reported on the rpl.rpl_checksum_cache test. The error reported was related to the binary log sender doing a heap-use-after-free on a given memory address. (Bug #78995, Bug #22109863)
- Replication: mysqldump has been updated to make it compatible with multi-source replication. Now when replication channels other than the default channel are found, mysqldump --dump-slave outputs a CHANGE MASTER TO statement for each replication channel. (Bug #78467, Bug #21855705)
- Replication: If a multi-threaded replication slave running with relay_log_recovery=1 stopped unexpectedly, during restart the relay log recovery process could fail. This was due to transaction inconsistencies not being filled, see Handling an Unexpected Halt of a Replication Slave. Prior to this fix, to recover from this situation required manually setting relay_log_recovery=0, starting the slave with START SLAVE UNTIL SQL_AFTER_MTS_GAPS to fix any transaction inconsistencies and then restarting the slave with relay_log_recovery=1. This process has now been automated, enabling relay log recovery of a multi-threaded slave upon restart automatically. (Bug #77496, Bug #21507981)
- Fedora builds now are configured using -DMYSQL_MAINTAINER_MODE=0 to silence GCC 6 warnings. (Bug #23274249)
- Allocation of a large number (2^20) of Performance Schema index statistic objects could cause a server exit. (Bug #23188107)
- If the keyring_okv plugin configuration directory was missing, attempts to rotate the InnoDB master key could cause a server exit. (Bug #23149683)
- INSERT with ON DUPLICATE KEY UPDATE and REPLACE on a table with a foreign key constraint defined failed with an incorrect “duplicate entry” error rather than a foreign key constraint violation error. (Bug #23135731)
- References: This issue is a regression of: Bug #78853, Bug #22037930.
- Contention in Performance Schema mutex instrumentation creation and destruction has been reduced, such that mutexes for which instruments are frequently created and destroyed are maintained in separate pages from those for which instruments are rarely created and destroyed. (Bug #22965826)
- With certain build options, an uninitialized variable in get_key_scans_params() could produce a compilation error. (Bug #22916059)
- Adding new tokens to the parser caused query digest values to change. (Bug #22906606)
- For debug builds, CONCAT_WS() could raise an assertion if there was nothing to append. (Bug #22888420)
- Fixed Valgrind warnings for failed LEAST() evaluations. (Bug #22883278)
- INET_NTOA() could cause a server exit when producing an error message. (Bug #22881810)
- References: This issue is a regression of: Bug #22042027.
- The my_write() call could cause a server exit if it attempted to check the current connection and the connection had been killed. (Bug #22867809)
- References: This issue is a regression of: Bug #21688407.
- Invoking Enterprise Encryption functions in multiple threads simultaneously could cause a server exit. (Bug #22839278)
- Setting log_syslog_tag to NULL could cause a server exit. (Bug #22834781)
- References: This issue is a regression of: Bug #22180046.
- If the expression for an indexed generated column contained an AND or OR operator, the optimizer could choose that index too often and create execution plans that produced incorrect results. (Bug #22810883)
- CREATE TABLE statements in mysqlpump output could be missing KEY clauses and would not load. (Bug #22726732)
- Attempting to use Enterprise Encryption functions after creating and dropping them could cause a server exit. (Bug #22669012)
- Setting sort_buffer_size to a very large value could cause some operations to fail and result in a server exit. (Bug #22594514)
- For an InnoDB table containing generated columns, using the table in a join could result in a server exit due to improper error checking. (Bug #22561845)
- REPLACE on a table with an indexed generated column could cause a server exit if the index prefix length was calculated incorrectly. (Bug #22445211)
- An UPDATE operation affecting a generated virtual BLOB column could cause a server exit. (Bug #22444212)
- SHOW CREATE TABLE for a table containing a generated column could cause a server exit or produce an Illegal mix of collations error. (Bug #22392268)
- On a slave server, replication of an UPDATE statement that updated an indexed BLOB virtual generated column of an InnoDB table could cause a server exit. (Bug #22241015)
- An assertion could be raised when a deadlock occurred due to a SELECT ... GROUP BY ... FOR UPDATE query executed using a Loose Index Scan. (Bug #22187476)
- mysqlpump could exit due to improper handling of error conditions in a dump thread. (Bug #22017120)
- Several potential buffer overflow issues were corrected. (Bug #21977380, Bug #23187436, Bug #23202778, Bug #23195370, Bug #23202699)
- If the CA certificate as given to the --ssl-ca option had an invalid path, yaSSL returned an error message different from OpenSSL. Now both return SSL connection error: SSL_CTX_set_default_verify_paths failed. (Bug #21920657)
- Installing MySQL from a yum or zypper repository resulted in /var/log/mysqld.log being created with incorrect user and group permissions. (Bug #21879694, Bug #78512)
- With show_compatibility_56=OFF, SHOW VARIABLES and SHOW STATUS statements having a WHERE clause did not work. (Bug #21783883)
- The audit_log plugin failed to abort the current operation when told to do so by a plugin handler for a MYSQL_AUDIT_PARSE_PREPARSE event. (Bug #21457956)
- Memory leaks reported by Valgrind for mysqlpump were fixed. (Bug #21237667)
- Some string functions returned one or a combination of their parameters as their result. If one of the parameters had a non-ASCII character set, the result string had the same character set, resulting in incorrect behavior when an ASCII string was expected. (Bug #18740222)
- On Fedora 24, upgrades using a Community MySQL Server RPM failed to replace an installed MariaDB Galera server due to a change in the MariaDB package. (Bug #81390, Bug #23273818)
- The mysql_read_defaults_options() function was missing a break statement, causing any option value for the --ssl-cipher option also to be applied to the --tls-version option, with unpredictable results. (Bug #81139, Bug #23129821)
- A DELETE from joined tables using a derived table in the WHERE clause failed if one of the joined tables was used in the derived table. (Bug #81014, Bug #23074801)
- Compiling of clients that used the MySQL C API could fail if they used an obsolete path to mysql.h or and did not set the include path. This was due to use of #include notation rather than #include "file_name" in internal MySQL header files, which have been adjusted. (Bug #80935, Bug #23047194)
- MySQL did not compile under Solaris 12 using Sun Studio. To correct this, instances of __attribute__ were changed to MY_ATTRIBUTE. (Bug #80748, Bug #22932576)
- The service_mysql_keyring.h and services.h header files misspelled mysql_keyring_service_st as mysql_keyring_file_service_st. (Bug #80688, Bug #22908232)
- The fix for Bug #79194 did not cover the eq_ref access method, with the result that left joins could return incorrect results.
- This bug fix has a very small negative performance effect such that it fails to cache an eq_ref-accessed row that is on the inner side of an outer join. Regular inner joins are not affected.
- References: This issue is a regression of: Bug #79194, Bug #22176604.
- The sanity() macro in strings/decimal.c produced Valgrind warnings due to reading uninitialized buffer contents. (Bug #80461, Bug #22782203, Bug #22839915)
- Starting the server with --initialize failed if the keyring_file_data system variable was also set at startup. To handle this, with --initialize the server no longer skips registration of plugins loaded with the --early-plugin-load option. (Bug #80451, Bug #22777039)
- Loading the Rewriter query rewrite plugin when there was no query_rewrite database resulted in Valgrind warnings. (Bug #80333, Bug #22710312)
- The optimizer transformed EXISTS (SELECT * ...) constructs to EXISTS (SELECT 1 ...) before all columns in the inner subquery had been resolved, which could result in rejection of valid queries that included a HAVING clause without GROUP BY in the subquery. (Bug #80231, Bug #22655856)
- For a server compiled with -DWITH_PERFSCHEMA_STORAGE_ENGINE=0, a memory leak could occur for buffered log messages used during server startup. (Bug #80089, Bug #22578574)
- A query could return incorrect results under these conditions: A column with a default value contained NULL; SELECT DISTINCT or a GROUP BY clause was used and the column containing the NULL value was part of the select list; an InnoDB temporary table was used during query processing. (Bug #79591, Bug #22343910)
- A null pointer dereference of a parser structure could occur during stored procedure name validation. (Bug #79396, Bug #22286421)
- Database initialization failed during installation using Ubuntu 15.10 packages. (Bug #79377, Bug #22252900)
- mysql_upgrade failed to widen the User and Proxied_user columns in the mysql.proxies_priv system table from 16 to 32 characters. (Bug #78254, Bug #21762656)
- Failure of UNINSTALL PLUGIN could lead to inaccurate or confusing errors for subsequent INSTALL PLUGIN operations. (Bug #74977, Bug #20085672)
- mysqld_multi displayed misleading error messages when it was unable to execute my_print_defaults. (Bug #74636, Bug #19920049)
- Previously, upgrading the server using an RPM package (including installation using yum) required upgrading the client package to the same MySQL version, which may be undesirable for some installations. This rule has been relaxed so that upgrading to a General Availability (GA) server version requires only that some GA client version be installed, which is less likely to require a client upgrade. (Bug #72230, Bug #18518216)
- mysqldump failed silently with no error message when it encountered an error while executing FLUSH LOGS. (Bug #71783, Bug #18284273)