MySQL 5.6 Release Notes
A known limitation of this release:
If you have InnoDB
tables with full-text
search indexes and you are upgrading from MySQL 5.6.10 to a
MySQL version up to and including MySQL 5.6.18, the server
will fail to start after the upgrade (Bug#72079). This bug is
fixed in MySQL 5.6.19. As a workaround, remove full-text
search indexes prior to upgrading and rebuild full-text search
indexes after the upgrade is completed.
Incompatible Change:
The
ERROR_FOR_DIVISION_BY_ZERO
,
NO_ZERO_DATE
, and
NO_ZERO_IN_DATE
SQL modes now
are deprecated and setting the sql_mode
value
to include any of them generates a warning. In MySQL 5.7, these
modes do nothing. Instead, their effects are included in the
effects of strict SQL mode
(STRICT_ALL_TABLES
or
STRICT_TRANS_TABLES
). The
motivation for the change in MySQL 5.7 is to reduce the number
of SQL modes with an effect dependent on strict mode and make
them part of strict mode itself.
To make advance preparation for an upgrade to MySQL 5.7, see SQL Mode Changes in MySQL 5.7. That discussion provides guidelines to assess whether your applications will be affected by the SQL mode changes in MySQL 5.7.
The msql2mysql, mysql_convert_table_format, mysql_find_rows, mysql_fix_extensions, mysql_setpermission, and mysqlaccess utilities are now deprecated and will be removed in MySQL 5.7. (Bug #27482, Bug #69012, Bug #69014, Bug #69015, Bug #69016, Bug #69017, Bug #11746603, Bug #16699248, Bug #16699279, Bug #16699284, Bug #16699317, Bug #18179576)
The IGNORE
clause for
ALTER TABLE
is now deprecated and
will be removed in a future version of MySQL. ALTER
IGNORE TABLE
causes problems for replication, prevents
online ALTER TABLE
for unique index creation,
and causes problems with foreign keys (rows removed in the
parent table).
Incompatible Change:
The AES_ENCRYPT()
and
AES_DECRYPT()
functions now
permit control of the block encryption mode and take an optional
initialization vector argument:
The new
block_encryption_mode
system variable controls the mode for block-based encryption
algorithms. Its default value is
aes-128-ecb
, which signifies encryption
using a key length of 128 bits and ECB mode.
An optional init_vector
argument
provides an initialization vector for encryption modes that
require it:
AES_ENCRYPT(str
,key_str
[,init_vector
]) AES_DECRYPT(crypt_str
,key_str
[,init_vector
])
A random string of bytes to use for the initialization
vector can be produced by calling the new
RANDOM_BYTES()
function.
For more information, see Encryption and Compression Functions.
These changes make statements that use
AES_ENCRYPT()
or
AES_DECRYPT()
unsafe for
statement-based replication and they cannot be stored in the
query cache. Queries that use
RANDOM_BYTES()
are unsafe for
statement-based replication and cannot be stored in the query
cache.
InnoDB:
Online DDL support is
extended to the following operations for regular and partitioned
InnoDB
tables:
ALTER TABLE ...
ENGINE=INNODB
(when run on an
InnoDB
table)
Online DDL support reduces table rebuild time and permits concurrent DML. See InnoDB and Online DDL.
(Bug #13975225)
Solaris:
On Solaris, mysql_config --libs now includes
-R
so that libraries can be found at runtime.
(Bug #18235669)/path/to/library
mysql_install_db provides a more informative diagnostic message when required Perl modules are missing. (Bug #69844, Bug #18187451)
Incompatible Change:
Old clients (older than MySQL 5.5.7) failed to parse
authentication data correctly if the server was started with the
--default-authentication-plugin=sha256_password
option.
As a result of this bug fix, MySQL 5.6.16 clients cannot
connect to a 5.6.17 server using an account that authenticates
with the sha256_password
plugin, nor can
5.6.17 clients connect to a 5.6.16 server. Similarly, MySQL
5.7.3 clients cannot connect to a 5.7.4 server using an
account that authenticates with the
sha256_password
plugin.
(Bug #17495562)
Important Change; InnoDB; Partitioning:
The
FLUSH
TABLES
statement's FOR EXPORT
option is now supported for partitioned
InnoDB
tables.
(Bug #16943907)
InnoDB:
Running a SELECT
on a partitioned table
caused a memory access violation in memcpy()
.
(Bug #18383840)
References: See also: Bug #18167648.
InnoDB:
For full-text queries, a failure to check that
num_token
is less than
max_proximity_item
could result in an
assertion.
(Bug #18233051)
InnoDB:
An invalid memmove
in
fts_query_fetch_document
would cause a
serious error.
(Bug #18229433)
InnoDB:
innodb_ft_result_cache_limit
now has a hardcoded maximum value of 4294967295 bytes or (2**32
-1). The maximum value was previously defined as the maximum
value of ulong
.
(Bug #18180057, Bug #71554)
InnoDB:
An UPDATE
resulted in a memory
access error in
lock_rec_other_trx_holds_expl
. The
transaction list (trx_sys->rw_trx_list
)
was traversed without acquiring the transaction subsystem mutex
(trx_sys->mutex
).
(Bug #18161853)
InnoDB:
InnoDB
failed to restore a corrupt first page
of a system tablespace data file from the doublewrite buffer,
resulting in a startup failure.
(Bug #18144349, Bug #18058884)
InnoDB: A regression introduced by Bug #14329288 would result in a performance degradation when a compressed table does not fit into memory. (Bug #18124788, Bug #71436)
References: This issue is a regression of: Bug #14329288.
InnoDB:
The maximum value for
innodb_thread_sleep_delay
is
now 1000000 microseconds. The previous maximum value (4294967295
microseconds on 32-bit and 18446744073709551615 microseconds on
64-bit) was unnecessarily large. Because the maximum value of
innodb_thread_sleep_delay
is
limited by the value set for
innodb_adaptive_max_sleep_delay
(when set to a nonzero value), the maximum value for
innodb_thread_sleep_delay
is
now the same as the maximum value for
innodb_adaptive_max_sleep_delay
.
(Bug #18117322)
InnoDB:
Attempting to uninstall the InnoDB
memcached plugin while the
InnoDB
memcached plugin is
still initializing would kill the InnoDB
memcached daemon thread. Uninstall should
wait until initialization is complete.
(Bug #18038948)
InnoDB: A full-text tokenizer thread would terminate with an incorrect error message. (Bug #18021306)
InnoDB: In debug builds, creating a unique index on a binary column, with input data containing duplicate keys, would cause an assertion. (Bug #18010711)
InnoDB:
The srv_monitor_thread
would crash in the
lock_print_info_summary()
function due to a
race condition between the srv_monitor_thread
and purge coordinator thread.
(Bug #17980590, Bug #70430)
InnoDB:
Attempting to add an invalid foreign key when foreign key
checking is disabled
(foreign_key_checks=0
) would
cause a serious error.
(Bug #17666774)
InnoDB:
For debug builds, the table rebuilding variant of online
ALTER TABLE
, when run on tables
with BLOB columns, would cause an assertion in the
row_log_table_apply_update
function. For
normal builds, a DB_PRODUCTION
error would be
returned.
(Bug #17661919)
InnoDB:
When creating a table there are a minimum of three separate
inserts on the mysql.innodb_index_stats
table. To improve CREATE TABLE
performance, there is now a single
COMMIT
operation instead of one
for each insert.
(Bug #17323202, Bug #70063)
InnoDB:
The server would halt with an assertion in
lock_rec_has_to_wait_in_queue(lock)
due to a
locking-related issue and a transaction being prematurely
removed from trx_sys->rw_trx_set
.
(Bug #17320977)
InnoDB:
Server shutdown would result in a hang with the following
message written to the error log: “[NOTE] InnoDB:
Waiting for purge thread to be suspended
.”
(Bug #16495065)
InnoDB:
InnoDB
failed to start when
innodb_data_file_path
specified
the data file size in kilobytes by appending
K
to the size value.
(Bug #16287752)
InnoDB: An insert buffer merge would cause an assertion error due to incorrectly handled ownership information for externally stored BLOBs.
InnoDB: Assertion failure in thread thread_num
in file ibuf0ibuf.cc line 4080
InnoDB: Failing assertion: rec_get_deleted_flag(rec, page_is_comp(page))
(Bug #14668683)
InnoDB:
Decreasing the
auto_increment_increment
value
would have no affect on the next auto-increment value.
(Bug #14049391, Bug #65225)
Partitioning:
When the
index_merge_intersection
flag
(enabled by default) or the
index_merge_union
flag was
enabled by the setting of the
optimizer_switch
system
variable, queries returned incorrect results when executed
against partitoned tables that used the
MyISAM
storage engine, as well as
partitioned InnoDB
tables that
lacked a primary key.
(Bug #18167648)
References: See also: Bug #16862316, Bug #17588348, Bug #17648468.
Replication:
The MASTER_SSL_CRL
and
MASTER_SSL_CRLPATH
options are not available
when using yaSSL; MySQL Replication now sets these to
NULL
automatically whenever yaSSL is enabled.
(Bug #18165937)
Replication:
Setting slave_parallel_workers
to 1 or greater and starting the slave caused the slave SQL
thread to use but not release memory until the slave was
restarted with STOP SLAVE
and
START SLAVE
.
(Bug #18001777, Bug #71197)
Replication:
When a slave was configured with replication filters and
--log-warnings=2
, every statement
which was filtered caused an entry to be written in the error
log. For busy servers which generated many statements to be
filtered, the result was that the error log could quickly grow
to many gigabytes in size. Now a throttle is used for such
errors, so that an error message is printed only once in a given
interval, saying that this particular error occurred a specific
number of times during that interval.
(Bug #17986385)
Replication:
SHOW SLAVE STATUS
used incorrect
values when reporting MASTER_SSL_CRL
and
MASTER_SSL_CRLPATH
.
(Bug #17772911, Bug #70866)
References: This issue is a regression of: Bug #11747191.
Replication:
Binary log events could be sent to slaves before they were
flushed to disk on the master, even when
sync_binlog
was set to 1. This
could lead to either of those of the following two issues when
the master was restarted following a crash of the operating
system:
Replication cannot continue because one or more slaves are requesting replicate events that do not exist on the master.
Data exists on one or more slaves, but not on the master.
Such problems are expected on less durable settings
(sync_binlog
not equal to 1), but it should
not happen when sync_binlog
is 1. To fix this
issue, a lock (LOCK_log
) is now held during
synchronization, and is released only after the binary events
are actually written to disk.
(Bug #17632285, Bug #70669)
Replication:
When running the slave with
slave_parallel_workers
at 1 or
greater, setting
--slave-skip-errors=all
caused
the error log to be filled with instances of the warning
Slave SQL: Could not execute Query event. Detailed
error: ;, Error_code: 0.
(Bug #17581990, Bug #68429)
References: See also: Bug #17986385.
Replication:
A number of possible state messages used as values for the
PROCESSLIST_STATE
column of the
threads
Performance Schema table
were longer than the width of the column (64 characters).
The long state messages are now silently truncated in order to avoid errors. This fix applies in MySQL 5.6 only; a permanent fix for the issue is made in MySQL 5.7 and later. (Bug #17319380)
Replication: The server did not handle correctly the insertion of a row larger than 4 GB when using row-based replication. (Bug #17081415)
Replication: When using row-based replication, an additional auto-increment column on the slave version of a table was not updated correctly; a zero was inserted instead. (Bug #17066269, Bug #69680)
Replication:
Statements involving the Performance Schema tables should not be
written to the binary log, because the content of these tables
is applicable only to a given MySQL Server instance, and may
differ greatly between different servers in a replication
topology. The database administrator should be able to configure
(INSERT
,
UPDATE
, or
DELETE
) or flush
(TRUNCATE TABLE
) performance
schema tables on a single server without affecting others.
However, when replicating from a MySQL 5.5 master to a MySQL 5.5
or later slave, warnings about unsafe statements updating
Performance Schema tables were elevated to errors. For MySQL 5.6
and later slaves, this prevented the simultaneous use of
performance_schema
and GTIDs (see
Replication with Global Transaction Identifiers).
This fix causes all updates on tables in the
performance_schema
database to be filtered on
the master and not replicated, regardless of the type of logging
that is in effect. Prior to this fix, statements using were
handled by being marked as unsafe for replication, which caused
warnings during execution; the statements were nonetheless
written to the binary log, regardless of the logging format in
effect.
Existing replication behavior for tables in the
INFORMATION_SCHEMA
database is not changed by
this fix.
For more information, see MySQL Performance Schema. (Bug #16814264)
References: See also: Bug #14741537, Bug #18259193.
Replication:
Modifying large amounts of data within a transaction can cause
the creation of temporary files. Such files are created when the
size of the data modified exceeds the size of the binary log
cache (max_binlog_cache_size
).
Previously, such files persisted until the client connection was
closed, which could allow them to grow until they exhausted all
available disk space in tmpdir
.
To prevent this from occurring, the size of a temporary file
created in this way in a given transaction is now reset to 0
when the transaction is committed or rolled back.
(Bug #15909788, Bug #18021493, Bug #66237)
Replication: The server checks to determine whether semisynchronous replication has been enabled without a lock, and if this is the case, it takes the lock and checks again. If semisynchronous replication was disabled after the first but prior to the second one, this could cause the server to fail. (Bug #14511533, Bug #66411)
References: See also: Bug #17920923.
Replication: Semisynchronous replication became very slow if there were many dump threads (such as from mysqlbinlog or slave I/O connections) working at the same time. It was also found that semisynchronous master plugin functions were called even when the dump connections did not support semisynchronous replication, which led to locking of the plugin lock as well as wasting time on necessary code.
After this fix, non-semisynchronous dump threads no longer call semisynchronous master functions to observe binary events. (Bug #70218, Bug #17434690)
Microsoft Windows:
On Microsoft Windows, the rw-lock backup implementation for the
my_atomic_*
functions was always used. Now,
the native Microsoft Windows implementation is used, where
available.
(Bug #18054042)
mysql_install_db could hang while reading
/dev/random
to generate a random
root
password.
(Bug #18395378)
While printing the server version, the mysql client did not check for buffer overflow in a string variable. (Bug #18186103)
Compilation failed if MySQL was configured with
CFLAGS
set to include a
-Werror
option with an argument.
(Bug #18173037)
A shared libmysqld
embedded server library
was not built on Linux. A new
WITH_EMBEDDED_SHARED_LIBRARY
CMake option now makes this possible.
(Bug #18123048, Bug #16430656, Bug #68559)
Building MySQL from source on Windows using Visual Studio 2008 failed with an identifier not found error due to a regression introduced by the patch for Bug#16249481. (Bug #18057449)
References: This issue is a regression of: Bug #16249481.
When tables are reopened from the table cache and the current thread is not instrumented for the Performance Schema, a table handle was unnecessarily instrumented. (Bug #18047865)
The SUM_SORT_MERGE_PASSES
column value in the
events_statements_summary_by_digest
Performance Schema table was calculated incorrectly.
(Bug #17938255)
If the
events_statements_summary_by_digest
Performance Schema table was full when a statement with a new
digest was found, the
Performance_schema_digest_lost
status variable was not incremented.
(Bug #17935314)
The audit log plugin could cause a server exit during log file rotation operations when there were many operations happening for multiple connections. (Bug #17930339)
The optimizer could push down a condition when the index did not have the key part present in the condition. (Bug #17814492)
Contraction information in a collation could be mishandled, resulting in incorrect decisions about whether a character is part of a contraction, and miscalculation of contraction weights. (Bug #17760379)
DROP TRIGGER
succeeded even with
the read_only
system variable
enabled.
(Bug #17503460)
If used to process a prepared
CALL
statement for a stored
procedure with OUT
or
INOUT
parameters,
mysql_stmt_store_result()
did
not properly set the flags required to retrieve all the result
sets.
(Bug #14492429, Bug #17849978)
Aggregating the results of a subquery in the
FROM
clause could produce incorrect results.
(Bug #71244, Bug #18014565)
A query that creates a temporary table to find distinct values and has a constant value in the projected list could produce incorrect results. (Bug #70657, Bug #17634335)
When run by root
, mysqld --help
--verbose exited with a nonzero error code after
displaying the help message.
(Bug #70058, Bug #17324415)
A deadlock error occurring during subquery execution could cause an assertion to be raised. (Bug #69969, Bug #17307201)
A temporal literal string without delimiters and more than 14
digits was validated as a TIMESTAMP/DATETIME
value with a two-digit precision fractional seconds part. But
fractional seconds should always be separated from other parts
of a time by a decimal point.
(Bug #69714, Bug #17080703)
For system variables that take a string value,
SET
statements permitted an unquoted value,
but values that contained dots were parsed incorrectly and only
part of the value was assigned. For example, SET GLOBAL
slow_query_log_file = my_slow.log
assigned the value
my_slow
. Now such values must be quoted or an
error occurs.
(Bug #69703, Bug #17075846)
The mysqladmin,
mysqlbinlog, mysqlcheck,
mysqldump, mysqlimport,
mysqlslap, and mysqlshow
programs now support a --secure-auth
option
that prevents sending passwords to the server in old (pre-4.1)
format. This option is enabled by default; use
--skip-secure-auth
to disable it.
(Bug #69051, Bug #16723046)
MySQL client programs from a Community Edition distribution
could not connect using SSL to a MySQL server from an Enterprise
Edition. This was due to a difference in certificate handling by
yaSSL and OpenSSL (used for Community and Enterprise,
respectively). OpenSSL expected a blank certificate to be sent
when not all of the --ssl-ca
,
--ssl-cert
, and --ssl-key
options were specified, and yaSSL did not do so. To resolve
this, yaSSL has been modified to send a blank certificate when
an option is missing.
(Bug #68788, Bug #16715064)
Messages written by the server to the error log for LDML collation definition problems were missing the collation name. (Bug #68144, Bug #16204175)
On Windows, mysql_install_db.pl could be run
only from within the bin
directory under
the installation directory.
(Bug #42421, Bug #11751526)