Author: Ben Kersten
Revision Date: 2010-10-03 22:59:26 UTC

Fixed bug where query status is not set to error when error packet is preceded by field header packets.

Author: Eric Herman
Revision Date: 2009-09-16 13:59:52 UTC

add some OS X install notes and notes for referencing MYSQL

Author: Eric Herman
Revision Date: 2009-09-16 12:20:47 UTC

fix for BUG#47370 'Proxy does not log what port it is listening on or proxying to'
 http://bugs.mysql.com/bug.php?id=47370

Author: Jan Kneschke
Revision Date: 2012-07-03 11:28:28 UTC

removed the _unlocked() suffix

  - all functions need to be locked with a mutex
  - chassis_component_registry_lock()/_unlock()

Author: <email address hidden>
Revision Date: 2010-09-30 13:25:33 UTC

added a test for chassis_fdlimit_set() and check that we don't try to set values what would
be interpreted wrongly

Author: <email address hidden>
Revision Date: 2010-09-17 14:24:37 UTC

replaced g_error() and g_assert() by proper error-handling and a debug-message

Author: Jan Kneschke
Revision Date: 2012-07-03 11:40:07 UTC

updated _register() parameters after rebase

Author: <email address hidden>
Revision Date: 2010-05-26 13:45:02 UTC

added missing Makefile.am for the master-plugin

Author: Jan Kneschke
Revision Date: 2012-06-12 11:29:26 UTC

fixed build on win32

Author: Kay Roepke
Revision Date: 2010-01-06 17:27:22 UTC

add support for Boehm GC

Author: Kay Roepke
Revision Date: 2010-01-08 16:40:11 UTC

add preliminary tcmalloc support with heap dump on SIGUSR1

Author: Kay Roepke
Revision Date: 2009-11-27 17:51:28 UTC

another try with non-portable legacy api

Author: Lachlan Mulcahy
Revision Date: 2009-11-24 02:16:55 UTC

Updated parameter name to be min_recheck_time instead of min_recheck_timeout, since it is a throttle, not a timeout specifically.

Author: Miguel Araújo
Revision Date: 2012-07-27 20:40:25 UTC

- fixed wrong plugin stats structure handling:

 - removed plugin_global_stats variable and unnecessary "global scope" functions;
       - added pointer on chassis_plugin_config that points to chassis_plugin_local_stats_t of chassis_plugin structure;
       - now we can access each plugin local stats through chassis_plugin_config;

- code refactoring on setting and getting statistic;
- fixed interface incorrect behaviour on some cases;

Author: Monty Taylor
Revision Date: 2007-08-06 23:47:48 UTC

* Added ppa suffix.

Author: Monty Taylor
Revision Date: 2007-08-06 23:42:50 UTC

* Updated debian/copyright file to point to versioned GPL.

Author: Jan Kneschke
Revision Date: 2012-05-25 06:50:13 UTC

removed section about Centos 4.3 and replaced fink by macports

Author: Matteo Nastasi
Revision Date: 2012-02-06 10:45:44 UTC

fixed var name for lua libs search

Author: Roger Booth
Revision Date: 2010-01-11 02:46:16 UTC

Fix for Proxy parser.lua MySQL Bug #36151

Author: Michael Schuster
Revision Date: 2010-04-23 09:29:39 UTC

codereview comments; minor fixes

Author: Michael Schuster
Revision Date: 2009-11-23 12:54:01 UTC

cscope

Author: Michael Schuster
Revision Date: 2009-12-09 14:41:41 UTC

fix for minor items

Author: Michael Schuster
Revision Date: 2010-02-26 08:14:18 UTC

fix copyright (year)

Author: Michael Schuster
Revision Date: 2010-05-10 12:11:43 UTC

mismerge fixed

Author: Michael Schuster
Revision Date: 2009-12-03 10:44:46 UTC

more review comments

Author: Stas Sușcov
Revision Date: 2012-08-09 15:50:27 UTC

Remove README.TESTS from debian/docs.

Author: vinchen
Revision Date: 2012-02-15 10:09:43 UTC

fix the stuck problem when the clients' concurrency too much

  While doing stress testing, 5000 connections execute a short query through proxy constantly，the proxy would have no response after a while. If strace the proxy process, and all the event threads are always logged like this :
      sendto(4, ".", 1, 0, NULL, 0 <unfinished ...>

  The problem come from the function chassis_event_add, and there are too much event when the clients' concurrency too much.
  Now I modified the implement of chassis_event_add by using random num to determin the handle thread's event_base. Moreover, the event base of libevent must be thread-safe.
  So the proxy should depend on libevent which version no less than 2.0(event base is thread safe https://raw.github.com/libevent/libevent/patches-2.0/whatsnew-2.0.txt).

  The problem is critical, and see also in
 http://bugs.mysql.com/bug.php?id=63908&error=no
 http://forums.mysql.com/read.php?146,433299,433299#msg-433299

Author: Diego Medina
Revision Date: 2009-02-07 03:59:23 UTC

* Now master_ssl_verify_server_cert is nil in LUA land for a master.info file from MySQL Server 4.1 and 5.0

Author: Diego Medina
Revision Date: 2009-02-06 16:30:28 UTC

* Finished fixing all the tutorial files

Author: ramon
Revision Date: 2009-04-06 12:00:28 UTC

Added keepalive and keepalive options to client sockets

We're facing a problem were at night when traffic is low the loadbalancer kills any open connections
due to timeout, however both client and server still think this connection is alive.
In the morning, all connections become more active fast, resulting in a mass of new client connections
while the old ones still have open sockets on the mysql-proxyed server.
This results in out-of-socket memory errors and a hungup server.

The code adds keepalive options to the client socket which results in sending a keepalive packet
every 2 minutes. This will ensure that the loadbalancer does not kill the connection

Author: trickie
Revision Date: 2009-07-16 11:19:54 UTC

add funnel lifecycle msc diagram

Author: Joshua Zhu
Revision Date: 2009-02-27 15:12:41 UTC

fixed: parse_value() may return false

Author: Mike Lischke
Revision Date: 2009-02-26 12:53:02 UTC

TAG:bundle-5.0-r17-win32

Author: Mats Kindahl
Revision Date: 2011-08-01 19:04:09 UTC

Patch to fix issues related to parsing transport URLs.
- Added unit tests for create_transport and
  fixed problems discovered.
- Rewrote create_transport:
  - Removed restriction to 20 characters file name.
  - Removed unnecessary dynamic memory allocation
  - Added default 3306 for port
- Fixed TCP URL parsing:
  - Empty user or hostname was accepted, which
    should not be accepted.
- Fixed file URL parsing:
  - Disallowed user and host instead of treating
    them as part of the file name.
  - File URLs according to spec accept host part,
    but we do not accept that currently.
- Binlog_file_driver::get_position() did not return
  file name correctly.
- Binlog_tcp_driver constructor initialized members
  twice. Moved initialization to initializer list.

Author: Mats Kindahl
Revision Date: 2013-04-20 09:49:19 UTC

Releasing version 0.1.0

Author: Mats Kindahl
Revision Date: 2013-04-20 09:49:19 UTC

Releasing version 0.1.0

Author: Giuseppe Maxia
Revision Date: 2008-08-31 09:51:28 UTC

- improved granularity of slave status

Author: Giuseppe Maxia
Revision Date: 2015-08-22 13:59:35 UTC

Last commit. This repo has migrated to GitHub

Author: Colin Charles
Revision Date: 2011-05-09 10:22:06 UTC

Remove aria* from clear scripts

Author: Diego Medina
Revision Date: 2010-04-16 11:06:13 UTC

* Use the agent to generate UUIDs instead of relaying on uuidgen (which is not on Solaris)

Author: Daniël van Eeden
Revision Date: 2013-10-23 15:25:03 UTC

Add mysql_config_editor to prevent errors like this:
$ ./my sql_config_editor print --all
/home/some_user/opt/mysql/5.6.14/bin/mysql_config_editor: unknown variable 'defaults-file=/home/some_user/sandboxes/msb_5_6_14/my.sandbox.cnf'

Author: Andrew Garner
Revision Date: 2011-03-25 15:25:37 UTC

fix bug #606206 - make_sandbox_from_installed no checking lib64/ paths

Author: Padraig O'Sullivan
Revision Date: 2010-02-16 22:14:19 UTC

Corrected issue with creating sandbox from 5.1 source trees.

Author: Daniel Nichter
Revision Date: 2010-07-18 23:18:02 UTC

Add options --package_name and --unpack_tarball to make_sandbox to support any package (MariaDB, Percona Server, etc.). Fix minor typos in README.

Author: Darren L. Oldag
Revision Date: 2012-10-03 15:18:27 UTC

merge to head of lp:mysql-sandbox, picking up fix for mysql 5.6.7 sandbox 'host' table issue

Author: Justin Tolmer
Revision Date: 2009-08-10 16:33:06 UTC

Fix group_id load/recovery when not running any transactional storage engines.

Author: Laurynas Biveinis
Revision Date: 2013-05-14 08:13:32 UTC

Port CVE-2012-5615 and CVE-2012-5627 patches from Percona Server 5.5

Author: Laurynas Biveinis
Revision Date: 2013-05-14 09:28:32 UTC

 Port CVE-2012-5615 and CVE-2012-5627 from Percona Server 5.6

Author: Tor Didriksen
Revision Date: 2013-11-01 15:39:19 UTC

Bug#17617945 BUFFER OVERFLOW IN GET_MERGE_MANY_BUFFS_COST WITH SMALL SORT_BUFFER_SIZE

get_cost_calc_buff_size() could return wrong value for the size of imerge_cost_buff.

Author: Balasubramanian Kandasamy
Revision Date: 2014-11-04 07:30:23 UTC

 Added sles11 repo packages

Author: Thirunarayanan B
Revision Date: 2014-11-21 05:20:02 UTC

Bug #19183565 CREATE DYNAMIC INNODB_TMPDIR VARIABLE TO CONTROL
  WHERE INNODB WRITES TEMP FILES
  - Reverting the patch.

Author: Jonas Oreland
Revision Date: 2012-03-09 13:24:50 UTC

ndb - add mapping table for MaxNoOfExecutionThreads to thread-types

Author: Ole John Aske
Revision Date: 2015-01-21 08:37:53 UTC

Fix regression introduced by fix for bug#19524096.

That fix caused ndb_global_schema_lock_error to fail as mysqld
ended up in a deadlock between TransporterFacade and ClusterMgr mutex
in ClusterMgr::is_cluster_completely_unavailable(). There
likely are other deadlock scenarios also.

The fix is to remove the Guard locking clusterMgrThreadMutex.
The rational and (lack of) correctness for this is discussed in
a mail pasted below.

There are also a few small improvements to ::is_cluster_completely_unavailable()

1. Dont copy entire 'trp_node' struct but read it by refference instead.

2. Replace usage of 'm_impl->m_transporter_facade' with 'getTransporter()'
   as this is the pattern used elsewhere in this code block.

--------------- Pasted mail with some background ----------------

Subject: (Dead-) locking of ClusterMgr::theNodes[] structures

Hi

Writing this as a note to myself and others after having analyzed
the failure of ndb_global_schema_lock_error.test. That test
timed out as mysqld ended up in a deadlock between
ClusterMgr::clusterMgrThreadMutex and TransporterFacade::theMutexPtr

ClusterMgr maintains node state & info in theNodes[]. From external
components, this info is access through ClusterMgr::getNodeInfo().
theNodes[] are only updated from within ClustMgr, all external access
is read only.

Updates to theNodes[] are partly done from withing several ClustMgr::exec<foo>
methods, and partly from ClusterMgr::reportConnected() / ::reportDisconnected().
All updates seems to be done with ClusterMgr::clusterMgrThreadMutex locked.

Several ClusterMgr methods are available for inspecting node status
from other components, these all use information from theNodes[].
Some of the more commonly used of these methods are:

 - TransporterFacade::get_node_alive(n) (Reads 'theClusterMgr->getNodeInfo(n)')
 - TransporterFacade::get_an_alive_node() (Implemented on top of ::get_node_alive(n))
 - NdbImpl::get_node_stopping(n) (Reads 'theClusterMgr->getNodeInfo(n)')
 - NdbImpl::get_node_alive(n) (Reads 'theClusterMgr->getNodeInfo(n)')
 - NdbImpl::get<foo> ...A lot more node state getters....

The locking schema used to provide atomicity of theNodes[] for the above
methods are .... mixed, and not really defined at all as far as I can tell.
Some examples:

- NdbDictInterface::dictSignal(), NdbDictInterface::forceGCPWait() & NdbDictInterface::listObjects():
  Before calling get_node_alive() / ::get_an_alive_node(), a PollGuard is set.
  PollGuard calls trp_client::start_poll() which is different pre/post 7.3:
  - Pre 7.3, trp_client::start_poll
              -> TransporterFacade::start_poll -
                -> lock TransporterFacade mutex (a global mutex)

   - 7.3 -> trp_client::start_poll, lock trp_client m_mutex.
              -> TransporterFacade::start_poll ...no locking, and mutex gone in this version

 Observations: There are no locking of ClusterMgr::clusterMgrThreadMutex here,
               neither pre/post 7.3 .

- Ndb_cluster_connection::wait_until_ready(),
  Ndb_cluster_connection_impl::get_next_alive_node()
  Ndb_cluster_connection::get_no_ready():
  These all sets the TransporterFacadeFurthermore mutex.

- Ndb::sendRecSignal
  Sets a PollGuard as above, which either lock the TransporterFacade or
  the trp_client mutex

- Ndb::sendPrepTrans()
  Documents in comments that TransporterFacade mutex should be locked prior to call

So this has become a total mess. It might seem like that it prior
to 7.3 was the intention that TransporterFacade mutex should be held
when accessing theNodes[], or any methods that access it itself.
After 7.3 a mix of TransporterFacade and Trp_client mutex is effectively used

Updating looks like it sets the ClusterMgr mutex to protect these, which will
of course not work as the reader doesnt set this mutex. However, it could be
that all updates happens at places where it is called from the TransporterFacade.
Here we *used to* hold the TransporterFacade mutex prior to 7.3, which would make
some sense. This all needs more investigation .... and some documentation in the code...
In the current state there certainly are no effective concurrency protection of
the node state info in 7.3+ , It could be that it work in 7.1 & 7.2

On top of this the fix for bug#19524096 introduced
ClusterMgr::is_cluster_completely_unavailable() which is also based on
the nodes status available in theNodes[]. Probably in good faith,
backed by that updates of theNodes[] was protected with clusterMgrThreadMutex,
that method grabs that lock before accessing theNodes[] info.

Actually this method is *the only* node state getters which
does any explicit locking. ... and it was doomed to fail as this
was completely untested territory. See other mail about how
it could deadlock with the TranporterFacade mutex.

Sidenote: any other node state getters attempting to follow the
same locking pattern had likely deadlocked the same way.

::is_cluster_completely_unavailable() is called from within code
which also calls ::get_node_alive() and ::get_an_alive_node(),
without using any locking protection for these. Based on this I will
propose a patch for the bug#19524096 regression, which simply
removes the mutex locking from ::is_cluster_completely_unavailable().

This will not be entirely kosher based on how the shared node state
structs should have been mutex protected. However, based on my discussion
above, there are already so many violations in this area that a single
more should not matter. A bigger effort should be taken to clean up this entirely.

Author: Mauritz Sundell
Revision Date: 2014-10-09 13:02:54 UTC

Cherrypicked

revision-id: mauritz.sundell@oracle.com-20141009124636-dg0th9bzvr27r1i7
parent: mauritz.sundell@oracle.com-20140930122950-gn1rl2yigc4s7ucu
committer: Mauritz Sundell <mauritz.sundell@oracle.com>
branch nick: mysql-7.1
timestamp: Thu 2014-10-09 14:46:36 +0200
message:
    Bug #19582807 MAKE SIGNAL DUMP IN TRACE FILES ALWAYS START FROM LATEST SIGNAL

    Removes a regression introduced with patch for the above bug.

    During crash dumps there could be a segmentation fault or most recent
    signals could be dump as old signals or not at all.

Author: orczhou
Revision Date: 2011-09-15 05:27:15 UTC

delete some extra space in the result file

Author: Daniel Volpato
Revision Date: 2008-08-28 16:18:14 UTC

Merge with main tree

Author: Davi Arnaut
Revision Date: 2013-02-06 21:34:47 UTC

Merge MySQL 5.5.30.

Author: Daniël van Eeden
Revision Date: 2013-05-24 21:15:05 UTC

Add test for --ignore-db-dir option

Author: Daniël van Eeden
Revision Date: 2011-04-18 11:14:33 UTC

Fix typo

Author: Daniël van Eeden
Revision Date: 2011-05-03 11:52:19 UTC

Change variables from log-err to error-log. Suggest the use of the
mysqld section instead of safe_mysqld. Also change safe_mysqld to
mysqld_safe.

Author: Daniël van Eeden
Revision Date: 2011-09-13 14:46:22 UTC

Add mysql.ufw for the Uncomplicated FireWall

Author: Daniël van Eeden
Revision Date: 2011-08-31 10:25:56 UTC

merge from upstream

Author: Andrew Gaul
Revision Date: 2014-01-16 23:10:38 UTC

Add missing whitespace to --general-log help text

Author: Andrew Gaul
Revision Date: 2014-01-23 23:25:23 UTC

Cooperate with log rotation when logging to a file.

Author: Hartmut Holzgraefe
Revision Date: 2012-06-07 19:07:56 UTC

Added a ndb_show_databases tool to print all databases known to cluster

This can be used to issue CREATE DATABASE IF NOT EXISTS statements
for all database to work around database creation auto detection
failures (e.g. when restoring metadata with ndb_restore, see
http://bugs.mysql.com/bug.php?id=46064 )

Author: He Zhenxing
Revision Date: 2010-03-05 02:35:49 UTC

Auto merge from 5.1 main

Author: He Zhenxing
Revision Date: 2009-04-29 09:25:18 UTC

BUG#44058 Possible semi-sync replication bugs (component part)

Semi-sync master component does not set the thread's proc info,
nor set the mysys_var->current_cond/mutex, which made KILL not
able to awake the thread that was waiting for the ACK, and didn't
know what the thread was waiting for.

This patch fixed the problem by calling enter_cond and exit_cond
accordingly in semi-sync master component before and after waiting
for ACK from slave.

Author: Igor Babaev
Revision Date: 2009-06-28 11:47:41 UTC

WL #4777: Support for persistent system tables to store basic statistics.
The main patch.

Author: Inaam Rana
Revision Date: 2014-08-06 21:14:55 UTC

Merge MySQL 5.5.39

Author: Inaam Rana
Revision Date: 2014-02-10 17:36:37 UTC

Merge MySQL 5.6.16

Author: Johannes Schlüter
Revision Date: 2009-01-06 13:09:31 UTC

Make gcc happy by using paranthesis

Author: Kevin Crane
Revision Date: 2015-02-05 22:44:20 UTC

Merge MySQL 5.5.41

Author: Kevin Crane
Revision Date: 2015-02-24 19:04:13 UTC

Merge MySQL 5.6.22

Author: Kostja Osipov
Revision Date: 2013-05-22 09:21:18 UTC

Update failing performance schema tests. Remove mention of LOCK_user_locks,
which is gone (no longer exists).

Author: Liang Guo
Revision Date: 2014-03-07 19:50:03 UTC

Merge MySQL 5.5.36

Author: Liang Guo
Revision Date: 2013-10-03 17:48:57 UTC

Merge MySQL 5.6.13

Author: Kelly Long
Revision Date: 2009-04-15 22:31:08 UTC

Kelly - Merge of mysql-6.0 into mysql-6.0-performance

Author: Kelly Long
Revision Date: 2009-05-04 16:57:10 UTC

Priority Scheduler that has some leakage to allow lower priority threads
 to make some progress.

Author: Mathias Gug
Revision Date: 2008-04-03 21:32:38 UTC

Drop 56-mysqlhotcopy-invalid-dbtable patch: fixed in 5.0.50 according to http://bugs.m/bug.php?id=27694.

Author: Monty Taylor
Revision Date: 2008-12-19 18:54:05 UTC

Removed mysql post-commit hooks.

Author: Bjorn Munch
Revision Date: 2014-03-19 10:32:19 UTC

Fix for bugs 18402580 and 18402999, as suggested for 18402580

  Instead of relying on $HOME, use Perl's getpwuid() to get home dir.

Author: Mauritz Sundell
Revision Date: 2014-10-09 13:04:27 UTC

Cherrypicked

revision-id: mauritz.sundell@oracle.com-20141009124636-dg0th9bzvr27r1i7
parent: mauritz.sundell@oracle.com-20140930122950-gn1rl2yigc4s7ucu
committer: Mauritz Sundell <mauritz.sundell@oracle.com>
branch nick: mysql-7.1
timestamp: Thu 2014-10-09 14:46:36 +0200
message:
    Bug #19582807 MAKE SIGNAL DUMP IN TRACE FILES ALWAYS START FROM LATEST SIGNAL

    Removes a regression introduced with patch for the above bug.

    During crash dumps there could be a segmentation fault or most recent
    signals could be dump as old signals or not at all.

Author: Padraig O'Sullivan
Revision Date: 2010-10-15 16:12:39 UTC

Forgot new method when embedded server is defined.