Release notes for 5.7.23-23 (57)

doc/source/conf.py

@@ -54,7 +54,7 @@
 # The short X.Y version.
 version = '5.7'
 # The full version, including alpha/beta/rc tags.
-release = '5.7.22-22'
+release = '5.7.23-23'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

doc/source/flexibility/max_binlog_files.rst

@@ -4,15 +4,43 @@
  Restricting the number of binlog files
 ========================================
 
-Maximum number of binlog files can now be restricted in |Percona Server| with :variable:`max_binlog_files`. When variable :variable:`max_binlog_files` is set to non-zero value, the server will remove the oldest binlog file(s) whenever their number exceeds the value of the variable. 
+Maximum number of binlog files can now be restricted in |Percona Server| with
+:variable:`max_binlog_files`. When variable :variable:`max_binlog_files` is set
+to non-zero value, the server will remove the oldest binlog file(s) whenever
+their number exceeds the value of the variable.
+
+This variable can be used with the existing :variable:`max_binlog_size` variable
+to limit the disk usage of the binlog files. If :variable:`max_binlog_size` is
+set to 1G and :variable:`max_binlog_files` to 20 this will limit the maximum
+size of the binlogs on disk to 20G. The actual size limit is not necessarily
+:variable:`max_binlog_size` * :variable:`max_binlog_files`. Server restart or
+``FLUSH LOGS`` will make the server start a new log file and thus resulting in
+log files that are not fully written in these cases limit will be lower.
 
-This variable can be used with the existing :variable:`max_binlog_size` variable to limit the disk usage of the binlog files. If :variable:`max_binlog_size` is set to 1G and :variable:`max_binlog_files` to 20 this will limit the maximum size of the binlogs on disk to 20G. The actual size limit is not necessarily :variable:`max_binlog_size` * :variable:`max_binlog_files`. Server restart or ``FLUSH LOGS`` will make the server start a new log file and thus resulting in log files that are not fully written in these cases limit will be lower. 
+Example
+=======
+
+Number of the binlog files before setting this variable :: 
+
+  $ ls -l mysql-bin.0* | wc -l
+  26
+
+Variable :variable:`max_binlog_files` is set to 20: ::
+
+  max_binlog_files = 20
+
+In order for new value to take effect ``FLUSH LOGS`` needs to be run. After that the number of binlog files is 20 :: 
+
+  $ ls -l mysql-bin.0* | wc -l
+  20
 
 Version Specific Information
 ============================
 
   * 5.7.10-1:
     Variable :variable:`max_binlog_files` ported from |Percona Server| 5.6.
+  * 5.7.23-23:
+    Variable :variable:`max_binlog_files` is deprecated and replaced with :variable:`binlog_space_limit`.
 
 System Variables
 ================
@@ -26,21 +54,23 @@ System Variables
      :vartype: ULONG
      :default: 0 (unlimited)
      :range: 0-102400
+
+.. variable:: binlog_space_limit
 
-Example
-=======
-
-Number of the binlog files before setting this variable :: 
-
-  $ ls -l mysql-bin.0* | wc -l
-  26
-
-Variable :variable:`max_binlog_files` is set to 20: ::
-
-  max_binlog_files = 20
-
-In order for new value to take effect ``FLUSH LOGS`` needs to be run. After that the number of binlog files is 20 :: 
+     :cli: Yes
+     :conf: Yes
+     :scope: Global
+     :dyn: No
+     :vartype: ULONG
+     :default: 0 (unlimited)
+     :range: 0-102400
+
+This option places an upper limit on the total size in bytes of all binary logs. A value of ``0`` means
+“no limit”. This is useful for a server host that has limited disk space.
 
-  $ ls -l mysql-bin.0* | wc -l
-  20
+When the limit is reached, oldest binary logs are purged until the total size is under the limit or only
+active log is remaining.
 
+.. note:: You should not set ``--binlog-space-limit`` to less or equal than the value of
+          ``--max-binlog-size`` because after the max-binlog-size limit will be reached, logs will be
+          rotated and immediately pruned by binlog-space-limit.

doc/source/management/data_at_rest_encryption.rst

@@ -103,6 +103,34 @@ number 13 set if tablespace is encrypted. This bit can be ckecked with
 System Variables
 ----------------
 
+.. variable:: innodb_sys_tablespace_encrypt
+
+  :version 5.7.23-23: Implemented
+  :cli: ``--innodb-sys-tablespace-encrypt``
+  :dyn: Yes
+  :scope: Global
+  :vartype: Boolean
+  :default: ``Off``	      
+
+When this variable is enabled, all data in the InnoDB system tablespace are
+encrypted.
+
+This feature is considered **ALPHA** quality.
+
+.. variable:: innodb_parallel_dblwr_encrypt
+
+  :version 5.7.23-23: Implemented
+  :cli: ``--innodb_parallel_dblwr_encrypt``
+  :dyn: Yes
+  :scope: Global
+  :vartype: Boolean
+  :default: ``Off``	      
+
+When this variable is enabled, all data in the parallel double write buffer are
+encrypted.
+
+This feature is considered **ALPHA** quality.
+
 .. variable:: innodb_temp_tablespace_encrypt
 
   :version 5.7.21-21: Implemented
@@ -112,7 +140,9 @@ System Variables
   :vartype: Boolean
   :default: ``Off``
 
-When this option is turned on, server starts to encrypt temporary tablespace
+This feature is considered **BETA** quality.
+
+When this option is turned on, the server starts to encrypt temporary tablespace
 and temporary |InnoDB| file-per-table tablespaces. The option does not force
 encryption of temporary tables which are currently opened, and it doesn't
 rebuild system temporary tablespace to encrypt data which are already written.
@@ -124,8 +154,8 @@ Turning this option off at runtime makes server to create all subsequent
 temporary file-per-table tablespaces unencrypted, but does not turn off
 encryption of system temporary tablespace.
 
-.. note:: To use this option, keyring plugin must be loaded, otherwise server
-   will give error message and refuse to create new temporary tables.
+.. note:: To use this option, the keyring plugin must be loaded, otherwise
+   server will give an error message and refuse to create new temporary tables.
 
 .. variable:: innodb_encrypt_tables
 
@@ -136,6 +166,8 @@ encryption of system temporary tablespace.
   :vartype: Text
   :default: ``OFF``
 
+This feature is considered **BETA** quality.
+
 This variable has 3 possible values. ``ON`` makes |InnoDB| tables encrypted by
 default. ``FORCE`` disables creation of unencrypted tables. ``OFF`` restores
 the like-before behavior.
@@ -195,6 +227,8 @@ A new feature, implemented since |Percona Server| :rn:`5.7.22-22`, is
 encryption of temporary files, triggered by the :variable:`encrypt-tmp-files`
 option.
 
+This feature is considered **BETA** quality.
+
 Temporary files are currently used in |Percona Server| for the following
 purposes:
 

doc/source/ps-variables.rst

@@ -77,6 +77,11 @@ System Variables
      - Yes
      - Global
      - No
+   * - :variable:`binlog_space_limit`
+     - Yes
+     - Yes
+     - Global
+     - Yes
    * - :variable:`csv_mode`
      - Yes
      - Yes
@@ -132,12 +137,12 @@ System Variables
      - Yes
      - Global
      - Yes
-  * - :variable:`innodb_encrypt_online_alter_logs`
+   * - :variable:`innodb_encrypt_online_alter_logs`
      - Yes
      - Yes
      - Global
      - Yes
-  * - :variable:`innodb_encrypt_tables`
+   * - :variable:`innodb_encrypt_tables`
      - Yes
      - Yes
      - Global

doc/source/release-notes/Percona-Server-5.7.22-22.rst

@@ -18,7 +18,6 @@ New Features
 * A new ``--encrypt-tmp-files`` option turns on encryption for the temporary
   files which  |Percona Server| may create on disk for filesort, binary log
   transactional caches and Group Replication caches.
-.
 
 Bugs Fixed
 ==========
@@ -28,7 +27,15 @@ Bugs Fixed
   its nature a counter and allowing only an increase, could return to its
   previous value. Bug fixed :psbug:`3951` (upstream :mysqlbug:`90351`).
 
-* NUMA support was improved in |Percona Server|, reverting upstream implementation back to the original one, due to upstream variant being less effective in memory allocation. Now `innodb_numa_interleave <http://dev.mysql.com/doc/refman/5.7/en/innodb-parameters.html#sysvar_innodb_numa_interleave>`_ variable not only enables NUMA interleave memory policy for the InnoDB buffer pool allocation, but forces NUMA interleaved allocation at the buffer pool initialization time. Bug fixed :psbug:`3967`.
+* NUMA support was improved in |Percona Server|, reverting upstream
+  implementation back to the original one, due to upstream variant
+  being less effective in memory allocation. Now
+  `innodb_numa_interleave
+  <http://dev.mysql.com/doc/refman/5.7/en/innodb-parameters.html#sysvar_innodb_numa_interleave>`_
+  variable not only enables NUMA interleave memory policy for the
+  InnoDB buffer pool allocation, but forces NUMA interleaved
+  allocation at the buffer pool initialization time. Bug fixed
+  :psbug:`3967`.
 
 * :variable:`audit_log_include_accounts` variable did not take effect if
   placed in ``my.cnf`` configuration file, while still working as intended if
@@ -60,22 +67,21 @@ MyRocks Changes and Fixes
   a full table scan on tables with unique secondary index. Bug fixed
   :psbug:`4495` (upstream `facebook/mysql-5.6#830 <https://github.com/facebook/mysql-5.6/issues/830>`_).
 
-
 Other Bugs Fixed
 ================
 
-* :psbug:`4451` "Implement better compression algo testing"
+* :psbug:`4451` \"Implement better compression algo testing\"
 
-* :psbug:`4469` "variable use out of scope bug in get_last_key test detected by
-  ASAN in clang 6"
+* :psbug:`4469` \"variable use out of scope bug in get_last_key test detected by
+  ASAN in clang 6\"
 
-* :psbug:`4470` "the cachetable-simple-pin-nonblocking-cheap test occasionally
-  fails due to a locking conflict with the cachetable evictor"
+* :psbug:`4470` \"the cachetable-simple-pin-nonblocking-cheap test occasionally
+  fails due to a locking conflict with the cachetable evictor\"
 
-* :psbug:`4488` "`-Werror` is always disabled for `innodb_memcached`"
+* :psbug:`4488` \"\`-Werror\` is always disabled for \`innodb_memcached\`\"
 
-* :psbug:`1114` "Assertion `inited == INDEX' failed"
+* :psbug:`1114` \"Assertion \`inited \=\= INDEX\' failed\"
 
-* :psbug:`1130` "RBR Replication with concurrent XA in READ-COMMITTED takes
-  supremum pseudo-records and breaks replication"
+* :psbug:`1130` \"RBR Replication with concurrent XA in READ-COMMITTED takes
+  supremum pseudo-records and breaks replication\"
 

doc/source/release-notes/Percona-Server-5.7.23-23.rst

@@ -0,0 +1,97 @@
+.. rn:: 5.7.23-23
+
+========================
+Percona Server |release|
+========================
+
+Percona is glad to announce the release of Percona Server |release| on
+September 12, 2018. Downloads are available `here
+<http://www.percona.com/downloads/Percona-Server-5.7/Percona-Server-5.7.23-23/>`_
+and from the :doc:`Percona Software Repositories </installation>`.
+
+This release is based on `MySQL 5.7.23
+<http://dev.mysql.com/doc/relnotes/mysql/5.7/en/news-5-7-23.html>`_
+and includes all the bug fixes in it. |Percona Server| |release| is
+now the current GA (Generally Available) release in the 5.7 series.
+
+All software developed by Percona is open-source and free.
+
+New Features
+============
+
+* The :variable:`max_binlog_files` variable is deprecated and replaced with
+  the :variable:`binlog_space_limit` variable. The behavior of
+  :variable:`binlog_space_limit` is consistent with the variable
+  :variable:`relay-log-space-limit` used for relay logs; both variables have the
+  same semantics. For more information, see :psbug:`275`.
+* Starting with |release|, it is possible to encrypt all data in the InnoDB
+  system tablespace and in the parallel double write buffer. A new variable
+  :variable:`innodb_sys_tablespace_encrypt` is introduced to encrypt the system
+  tablespace. This feature is considered **ALPHA** quality. The encryption of
+  the parallel double write buffer file is controlled by the variable
+  :variable:`innodb_parallel_dblwr_encrypt`. Both variables are ``OFF`` by
+  default. For more information, see :psbug:`3822`.
+* Changing ``rocksdb_update_cf_options`` returns any warnings and errors to the
+  client instead of printing them to the server error log. For more information,
+  see :psbug:`4258`.
+* :variable:`rocksdb_number_stat_computers` and 
+  :variable:`rocksdb_rate_limit_delay_millis` variables have been removed. For
+  more information, see :psbug:`4780`.
+* A number of new variables were introduced for |MyRocks|: 
+  :variable:`rocksdb_rows_filtered` to show the number of rows filtered out for
+  TTL in MyRocks tables, :variable:`rocksdb_bulk_load_allow_sk` to allow adding
+  secondary keys using the bulk loading feature,
+  :variable:`rocksdb_error_on_suboptimal_collation` toggling warning or error
+  in case of an index creation on a char field where the table has a sub-optimal
+  collation, :variable:`rocksdb_stats_recalc_rate` specifying the number of
+  indexes to recalculate per second,
+  :variable:`rocksdb_commit_time_batch_for_recovery` toggler of writing the
+  commit time write batch into the database,
+  and :variable:`rocksdb_write_policy` specifying when two-phase commit data are
+  actually written into the database.
+
+Bugs Fixed
+==========
+
+* The statement ``SELECT...ORDER BY`` produced inconsistent results with the
+  ``euckr`` charset or ``euckr_bin`` collation. Bug fixed :psbug:`4513`
+  (upstream :mysqlbug:`91091`).
+* InnoDB statistics could incorrectly report zeros in the :ref:`slow query log
+  <slow_extended>`. Bug fixed :psbug:`3828`.
+* With the FIPS mode enabled and performance_schema=off, the instance crashed
+  when running the ``CREATE VIEW`` command. Bug fixed :psbug:`3840`.
+* The soft limit of the core file size was set incorrectly starting with PS
+  :rn:`5.7.21-20`. Bug fixed :psbug:`4479`.
+* The option ``innodb-optimize-keys`` could fail when a dumped table has two
+  columns such that the name of one of them contains the other as as a prefix and
+  is defined with the AUTO_INCREMENT attribute. Bug fixed :psbug:`4524`.
+* When :variable:`innodb_temp_tablespace_encrypt` was set to ``ON`` the ``CREATE TABLE``
+  command could ignore the value of the ``ENCRYPTION`` option. Bug fixed
+  :psbug:`4565`.
+* If ``FLUSH STATUS`` was run from a different session, a statement could be
+  counted twice in ``GLOBAL STATUS``. Bug fixed :psbug:`4570` (upstream
+  :mysqlbug:`91541`).
+* In some cases, it was not possible to set the :variable:`flush_caches`
+  variable on systems that use systemd. Bug fixed :psbug:`3796`.
+* A message in the |MyRocks| log file did not clearly inform whether fast CRC32
+  was supported. Bug fixed :psbug:`3988`.
+* ``mysqld`` could not be started on Ubuntu if the database recovery had taken
+  longer than ten minutes. Bug fixed :psbug:`4546` (upstream :mysqlbug:`91423`).
+* The ALTER TABLE command was slow when the number of dirty pages was high. Bug fixed
+  :psbug:`3702`.
+* Setting the global variable :variable:`version_suffix` to NULL could
+  lead to a server crash. Bug fixed :psbug:`4785`.
+
+Other Bugs Fixed
+================
+
+* :psbug:`4620` \"Enable encryption of temporary tablespace from foreground thread\"
+* :psbug:`4727` \"intrinsic temp table behaviour shouldn\'t depend on innodb_encrypt_tables\"
+* :psbug:`4046` \"Ship assert failure: \'res == 0\' (bulk loader)\"
+* :psbug:`3851` \"Percona Ver 5.6.39-83.1 Failing assertion: sym_node->table != NULL\"
+* :psbug:`4533` \"audit_log MTR tests should refer to include files without parent directories\"
+* :psbug:`4619` \"main.flush_read_lock fails with timeout in wait_condition.inc.\" 
+* :psbug:`4561` \"Read after free at Binlog_crypt_data::load_latest_binlog_key()\"
+* :psbug:`4587` \"ROCKSDB_INCLUDE_RFR macro in wrong file\"
+
+.. |release| replace:: 5.7.23-23

doc/source/release-notes/release-notes_index.rst

@@ -6,6 +6,7 @@
    :maxdepth: 1
    :glob:
 
+   Percona-Server-5.7.23-23
    Percona-Server-5.7.22-22
    Percona-Server-5.7.21-21
    Percona-Server-5.7.21-20