mysql主从复制5.6基于GID及多线程的复制笔记

mysql:数据库复制过滤
     #show grobal variables like 'binlog%';
注意：不能在主服务器上做过滤，二进制不完整性；它将危害到其他的数据库；在从库过滤会占用过多的带宽,还会复制过多的跟数据库无关的内容
主端:
    binlog-do-db:仅将指定数据库的相关修改操作记为二进制日志(白名单)
    binlog-ignore-db
从端:
    replicate-do-db=    #仅复制指定数据库的二进制日志文件
    replicate-ignore-db= #忽略那数据库
    replicate-do-table=
    replicate-ignore-table=
    replicate-wild-do-table=
    replicate-wild-ignore-table=
在从库中：
修改配置文件：vim /etc/my.cnf
    replicate-do-db = discuz
#重启mysql
mysql> show variables like 'rep%';#查看rep
mysql> show slave statusG #查看slave的状态
从库测试结果：
mysql> show databases;
+--------------------+
| Database           |
+--------------------+
| information_schema |
| discuz 
主库：创建一个数据库discuz；从库同步成功看上面的结果：
2.1.1.mysql5.6
数据库中有多个数据库时，多线程复制才有意义；
MySQL-5.6: GTID
     slave-parallel-workers=0
  0: 表示禁用多线程功能；


     MySQL 5.6引入的GTID(Global Transaction IDs)使得其复制功能的配置、监控及管理变得更加易于实现，且更加健壮。
要在MySQL 5.6中使用复制功能，其服务配置段[mysqld]中于少应该定义如下选项：
binlog-format：二进制日志的格式，有row、statement和mixed几种类型；
 需要注意的是：当设置隔离级别为READ-COMMITED必须设置二进制日志格式为ROW，现在MySQL官方认为STATEMENT这个已经不再适合继续使用；但mixed类型在默认的事务隔离级别下，可能会导致主从数据不一致； 
log-slave-updates、gtid-mode、enforce-gtid-consistency、report-port和report-host：用于启动GTID及满足附属的其它需求；
master-info-repository和relay-log-info-repository：启用此两项，可用于实现在崩溃时保证二进制及从服务器安全的功能；
sync-master-info：启用之可确保无信息丢失；
slave-paralles-workers：设定从服务器的SQL线程数；0表示关闭多线程复制功能；
binlog-checksum、master-verify-checksum和slave-sql-verify-checksum：启用复制有关的所有校验功能；
binlog-rows-query-log-events：启用之可用于在二进制日志记录事件相关的信息，可降低故障排除的复杂度； 
log-bin：启用二进制日志，这是保证复制功能的基本前提；
server-id：同一个复制拓扑中的所有服务器的id号必须惟一；
一、简单主从模式配置步骤 
1、配置主从节点的服务配置文件 
1.1、配置master节点： 
[mysqld] 
binlog-format=ROW 
log-bin=master-bin 
log-slave-updates=true 
gtid-mode=on   #开启gtid功能
enforce-gtid-consistency=true 
master-info-repository=TABLE 
relay-log-info-repository=TABLE 
sync-master-info=1 
slave-parallel-workers=2 
binlog-checksum=CRC32 
master-verify-checksum=1 
slave-sql-verify-checksum=1 
binlog-rows-query-log_events=1 
server-id=1 
report-port=3306 
port=3306 
datadir=/mydata/data 
socket=/tmp/mysql.sock 
report-host=master.magedu.com 


1.2、配置slave节点： 
[mysqld] 
binlog-format=ROW 
log-slave-updates=true 
gtid-mode=on 
enforce-gtid-consistency=true 
master-info-repository=TABLE 
relay-log-info-repository=TABLE 
sync-master-info=1 
slave-parallel-workers=2 
binlog-checksum=CRC32 
master-verify-checksum=1 
slave-sql-verify-checksum=1 
binlog-rows-query-log_events=1 
server-id=11 
report-port=3306 
port=3306 
log-bin=mysql-bin.log 
datadir=/mydata/data 
socket=/tmp/mysql.sock 
report-host=slave.magedu.com 


2、创建复制用户 
mysql> GRANT REPLICATION SLAVE ON *.* TO repluser@172.16.100.7 IDENTIFIED BY 'replpass'; 
说明：172.16.100.7是从节点服务器；如果想一次性授权更多的节点，可以自行根据需要修改； 


3、为备节点提供初始数据集 
锁定主表，备份主节点上的数据，将其还原至从节点；如果没有启用GTID，在备份时需要在master上使用show master status命令查看二进制日志文件名称及事件位置，以便后面启动slave节点时使用。 


4、启动从节点的复制线程 
如果启用了GTID功能，则使用如下命令： 
mysql> CHANGE MASTER TO MASTER_HOST='master.magedu.com', MASTER_USER='repluser', MASTER_PASSWORD='replpass', MASTER_AUTO_POSITION=1; 


没启用GTID，需要使用如下命令： 
slave> CHANGE MASTER TO MASTER_HOST='172.16.100.6', 
-> MASTER_USER='repluser', 
-> MASTER_PASSWORD='replpass', 
-> MASTER_LOG_FILE='master-bin.000003', 
-> MASTER_LOG_POS=1174; 


二、半同步复制 
1、分别在主从节点上安装相关的插件 
master> INSTALL PLUGIN rpl_semi_sync_master SONAME 'semisync_master.so'; 
slave> INSTALL PLUGIN rpl_semi_sync_slave SONAME 'semisync_slave.so'; 
2、启用半同步复制 
在master上的配置文件中，添加 
rpl_semi_sync_master_enabled=ON 
在至少一个slave节点的配置文件中添加 
rpl_semi_sync_slave_enabled=ON 
而后重新启动mysql服务即可生效。 
或者，也可以mysql服务上动态启动其相关功能： 
master> SET GLOBAL rpl_semi_sync_master_enabled = ON; 
slave> SET GLOBAL rpl_semi_sync_slave_enabled = ON; 
slave> STOP SLAVE IO_THREAD; START SLAVE IO_THREAD; 
3、确认半同步功能已经启用 
master> CREATE DATABASE magedudb; 
master> SHOW STATUS LIKE 'Rpl_semi_sync_master_yes_tx'; 
slave> SHOW DATABASES; 
三、MySQL Proxy 
1、源码安装时，MySQL proxy的依赖关系： 
libevent 1.x or higher (1.3b or later is preferred). 
lua 5.1.x or higher. 
glib2 2.6.0 or higher. 
pkg-config. 
libtool 1.5 or higher. 
MySQL 5.0.x or higher developer files. 
2、安装 
# tar zxf mysql-proxy-0.8.2.tar.gz 
# cd mysql-proxy-0.8.2 
# ./configure 
# make 


# make check 
如果管理员有密码，上面的步骤则需要使用如下格式进行： 
# MYSQL_PASSWORD=root_pwd make check 


# make install 
默认情况下， mysql-proxy安装在/usr/local/sbin/mysql-proxy，而Lua示例脚本安装在/usr/local/share目录中。 


3、配置指令 
mysql proxy的各配置参数请参见官方文档，http://dev.mysql.com/doc/refman/5.6/en/mysql-proxy-configuration.html 


使用rpm包在rhel6上安装mysql-proxy-0.8.2，其会提供配置文件及服务脚本，但没有提供读写分享的脚本。 
/etc/sysconfig/mysql-proxy文件用于定义mysql-proxy的启动参数。 


ADMIN_USER – the user for the proxy's admin interface. You can leave the default admin user. 
ADMIN_PASSWORD – the password for the admin user in clear text. Change the default password for better security. 
ADMIN_LUA_SCRIPT – the admin script in the Lua programming language. Without this script the admin interface cannot work. You can leave the default value. 
PROXY_USER – the system user under which the proxy will work. By default it is mysql-proxy, and it's safe to leave it as is. 
PROXY_OPTIONS – proxy options such as logging level, plugins, and Lua scripts to be loaded. 


其中PROXY_OPTIONS是最常用的一个选项，用于定义mysql-proxy工作时的重要参数，例如： 
PROXY_OPTIONS="--daemon --log-level=info --log-use-syslog --plugins=proxy --plugins=admin --proxy-backend-addresses=192.168.1.102:3306 --proxy-read-only-backend-addresses=192.168.1.105:3306 --proxy-lua-script=/usr/lib/mysql-proxy/lua/proxy/rw-splitting.lua" 


四、安装配置mysql-proxy: 


4.1 下载所需要的版本，这里的系统平台为rhel6.4 32位系统，因此就以mysql-proxy-0.8.3-linux-glibc2.3-x86-32bit.tar.gz为例。 


# tar xf mysql-proxy-0.8.3-linux-glibc2.3-x86-32bit.tar.gz -C /usr/local 
# cd /usr/local 
# ln -sv mysql-proxy-0.8.3-linux-glibc2.3-x86-32bit mysql-proxy 


添加代理用户 
# useradd mysql-proxy 


4.2 为mysql-proxy提供SysV服务脚本，内容如下所示 


#!/bin/bash 
# 
# mysql-proxy This script starts and stops the mysql-proxy daemon 
# 
# chkconfig: - 78 30 
# processname: mysql-proxy 
# description: mysql-proxy is a proxy daemon for mysql 


# Source function library. 
. /etc/rc.d/init.d/functions 


prog="/usr/local/mysql-proxy/bin/mysql-proxy" 


# Source networking configuration. 
if [ -f /etc/sysconfig/network ]; then 
. /etc/sysconfig/network 
fi 


# Check that networking is up. 
[ ${NETWORKING} = "no" ] && exit 0 


# Set default mysql-proxy configuration. 
ADMIN_USER="admin" 
ADMIN_PASSWD="admin" 
ADMIN_LUA_SCRIPT="/usr/local/mysql-proxy/share/doc/mysql-proxy/admin.lua" 
PROXY_OPTIONS="--daemon" 
PROXY_PID=/var/run/mysql-proxy.pid 
PROXY_USER="mysql-proxy" 


# Source mysql-proxy configuration. 
if [ -f /etc/sysconfig/mysql-proxy ]; then 
. /etc/sysconfig/mysql-proxy 
fi 


RETVAL=0 


start() { 
echo -n $"Starting $prog: " 
daemon $prog $PROXY_OPTIONS --pid-file=$PROXY_PID --proxy-address="$PROXY_ADDRESS" --user=$PROXY_USER --admin-username="$ADMIN_USER" --admin-lua-script="$ADMIN_LUA_SCRIPT" --admin-password="$ADMIN_PASSWORD" 
RETVAL=$? 
echo 
if [ $RETVAL -eq 0 ]; then 
touch /var/lock/subsys/mysql-proxy 
fi 
} 


stop() { 
echo -n $"Stopping $prog: " 
killproc -p $PROXY_PID -d 3 $prog 
RETVAL=$? 
echo 
if [ $RETVAL -eq 0 ]; then 
rm -f /var/lock/subsys/mysql-proxy 
rm -f $PROXY_PID 
fi 
} 
# See how we were called. 
case "$1" in 
start) 
start 
;; 
stop) 
stop 
;; 
restart) 
stop 
start 
;; 
condrestart|try-restart) 
if status -p $PROXY_PIDFILE $prog >&/dev/null; then 
stop 
start 
fi 
;; 
status) 
status -p $PROXY_PID $prog 
;; 
*) 
echo "Usage: $0 {start|stop|restart|reload|status|condrestart|try-restart}" 
RETVAL=1 
;; 
esac 
exit $RETVAL 
将上述内容保存为/etc/rc.d/init.d/mysql-proxy，给予执行权限，而后加入到服务列表。 
# chmod +x /etc/rc.d/init.d/mysql-proxy 
# chkconfig --add mysql-proxy 

4.3 为服务脚本提供配置文件/etc/sysconfig/mysql-proxy，内容如下所示： 

# Options for mysql-proxy 
ADMIN_USER="admin" 
ADMIN_PASSWORD="admin" 
ADMIN_ADDRESS="" 
ADMIN_LUA_SCRIPT="/usr/local/mysql-proxy/share/doc/mysql-proxy/admin.lua" 
PROXY_ADDRESS="" 
PROXY_USER="mysql-proxy" 
PROXY_OPTIONS="--daemon --log-level=info --log-use-syslog" 

其中最后一行，需要按实际场景进行修改，例如： 
PROXY_OPTIONS="--daemon --log-level=info --log-use-syslog --plugins=proxy --plugins=admin --proxy-backend-addresses=172.16.100.6:3306 --proxy-read-only-backend-addresses=172.16.100.7:3306 --proxy-lua-script=/usr/local/mysql-proxy/share/doc/mysql-proxy/rw-splitting.lua" 

其中的proxy-backend-addresses选项和proxy-read-only-backend-addresses选项均可重复使用多次，以实现指定多个读写服务器或只读服务器。 
4.4 mysql-proxy的配置选项 

mysql-proxy的配置选项大致可分为帮助选项、管理选项、代理选项及应用程序选项几类，下面一起去介绍它们。 

--help 
--help-admin 
--help-proxy 
--help-all ———— 以上四个选项均用于获取帮助信息； 
--proxy-address=host:port ———— 代理服务监听的地址和端口； 
--admin-address=host:port ———— 管理模块监听的地址和端口； 
--proxy-backend-addresses=host:port ———— 后端mysql服务器的地址和端口； 
--proxy-read-only-backend-addresses=host:port ———— 后端只读mysql服务器的地址和端口； 
--proxy-lua-script=file_name ———— 完成mysql代理功能的Lua脚本； 
--daemon ———— 以守护进程模式启动mysql-proxy； 
--keepalive ———— 在mysql-proxy崩溃时尝试重启之； 
--log-file=/path/to/log_file_name ———— 日志文件名称； 
--log-level=level ———— 日志级别； 
--log-use-syslog ———— 基于syslog记录日志； 
--plugins=plugin,.. ———— 在mysql-proxy启动时加载的插件； 
--user=user_name ———— 运行mysql-proxy进程的用户； 
--defaults-file=/path/to/conf_file_name ———— 默认使用的配置文件路径；其配置段使用[mysql-proxy]标识； 
--proxy-skip-profiling ———— 禁用profile； 
--pid-file=/path/to/pid_file_name ———— 进程文件名； 

5、复制如下内容建立admin.lua文件，将其保存至/usr/local/mysql-proxy/share/doc/mysql-proxy/目录中。 
--[[ $%BEGINLICENSE%$ 
Copyright (c) 2007, 2012, Oracle and/or its affiliates. All rights reserved. 


This program is free software; you can redistribute it and/or 
modify it under the terms of the GNU General Public License as 
published by the Free Software Foundation; version 2 of the 
License. 

This program is distributed in the hope that it will be useful, 
but WITHOUT ANY WARRANTY; without even the implied warranty of 
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 
GNU General Public License for more details. 

You should have received a copy of the GNU General Public License 
along with this program; if not, write to the Free Software 
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 
02110-1301 USA 


$%ENDLICENSE%$ --]] 


function set_error(errmsg) 
proxy.response = { 
type = proxy.MYSQLD_PACKET_ERR, 
errmsg = errmsg or "error" 
} 
end 

function read_query(packet) 
if packet:byte() ~= proxy.COM_QUERY then 
set_error("[admin] we only handle text-based queries (COM_QUERY)") 
return proxy.PROXY_SEND_RESULT 
end 


local query = packet:sub(2) 

local rows = { } 
local fields = { } 

if query:lower() == "select * from backends" then 
fields = { 
{ name = "backend_ndx", 
type = proxy.MYSQL_TYPE_LONG }, 

{ name = "address", 
type = proxy.MYSQL_TYPE_STRING }, 
{ name = "state", 
type = proxy.MYSQL_TYPE_STRING }, 
{ name = "type", 
type = proxy.MYSQL_TYPE_STRING }, 
{ name = "uuid", 
type = proxy.MYSQL_TYPE_STRING }, 
{ name = "connected_clients", 
type = proxy.MYSQL_TYPE_LONG }, 
} 

for i = 1, #proxy.global.backends do 
local states = { 
"unknown", 
"up", 
"down" 
} 
local types = { 
"unknown", 
"rw", 
"ro" 
} 
local b = proxy.global.backends[i] 

rows[#rows + 1] = { 
i, 
b.dst.name, -- configured backend address 
states[b.state + 1], -- the C-id is pushed down starting at 0 
types[b.type + 1], -- the C-id is pushed down starting at 0 
b.uuid, -- the MySQL Server's UUID if it is managed 
b.connected_clients -- currently connected clients 
} 
end 
elseif query:lower() == "select * from help" then 
fields = { 
{ name = "command", 
type = proxy.MYSQL_TYPE_STRING }, 
{ name = "description", 
type = proxy.MYSQL_TYPE_STRING }, 
} 
rows[#rows + 1] = { "SELECT * FROM help", "shows this help" } 
rows[#rows + 1] = { "SELECT * FROM backends", "lists the backends and their state" } 
else 
set_error("use 'SELECT * FROM help' to see the supported commands") 
return proxy.PROXY_SEND_RESULT 
end 

proxy.response = { 
type = proxy.MYSQLD_PACKET_OK, 
resultset = { 
fields = fields, 
rows = rows 
} 
} 
return proxy.PROXY_SEND_RESULT 
end 

6、测试 

6.1 管理功能测试 
# mysql -uadmin -padmin -h172.16.100.107 --port=4041 
Welcome to the MySQL monitor. Commands end with ; or g. 
Your MySQL connection id is 1 
Server version: 5.0.99-agent-admin 
Type 'help;' or 'h' for help. Type 'c' to clear the buffer. 

mysql> SELECT * FROM backends; 
+-------------+-------------------+-------+------+------+-------------------+ 
| backend_ndx | address | state | type | uuid | connected_clients | 
+-------------+-------------------+-------+------+------+-------------------+ 
| 1 | 172.16.100.6:3306 | up | rw | NULL | 0 | 
| 2 | 172.16.100.7:3306 | up | ro | NULL | 0 | 
+-------------+-------------------+-------+------+------+-------------------+ 
2 rows in set (0.00 sec) 
6.2 读写分离测试 
# mysql -uroot -pmagedu.com -h172.16.100.107 --port=4040 
下面的读写分享脚本是由mysql-proxy-0.8.3提供了，将其复制保存为/usr/lib/mysql-proxy/lua/proxy/rw-splitting.lua，就可以启动服务了。 
--[[ $%BEGINLICENSE%$ 
Copyright (c) 2007, 2012, Oracle and/or its affiliates. All rights reserved. 

This program is free software; you can redistribute it and/or 
modify it under the terms of the GNU General Public License as 
published by the Free Software Foundation; version 2 of the 
License. 

This program is distributed in the hope that it will be useful, 
but WITHOUT ANY WARRANTY; without even the implied warranty of 
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 
GNU General Public License for more details. 

You should have received a copy of the GNU General Public License 
along with this program; if not, write to the Free Software 
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 
02110-1301 USA 

$%ENDLICENSE%$ --]] 

---
-- a flexible statement based load balancer with connection pooling 
-- 
-- * build a connection pool of min_idle_connections for each backend and maintain 
-- its size 
-- * 
-- 
local commands = require("proxy.commands") 
local tokenizer = require("proxy.tokenizer") 
local lb = require("proxy.balance") 
local auto_config = require("proxy.auto-config") 

--- config 
-- 
-- connection pool 
if not proxy.global.config.rwsplit then 
proxy.global.config.rwsplit = { 
min_idle_connections = 4, 
max_idle_connections = 8, 

is_debug = false 
} 
end 

--- 
-- read/write splitting sends all non-transactional SELECTs to the slaves 
-- 
-- is_in_transaction tracks the state of the transactions 
local is_in_transaction = false 

-- if this was a SELECT SQL_CALC_FOUND_ROWS ... stay on the same connections 
local is_in_select_calc_found_rows = false 

--- 
-- get a connection to a backend 
-- 
-- as long as we don't have enough connections in the pool, create new connections 
-- 
function connect_server() 
local is_debug = proxy.global.config.rwsplit.is_debug 
-- make sure that we connect to each backend at least ones to 
-- keep the connections to the servers alive 
-- 
-- on read_query we can switch the backends again to another backend 


if is_debug then 
print() 
print("[connect_server] " .. proxy.connection.client.src.name) 
end 

local rw_ndx = 0 

-- init all backends 
for i = 1, #proxy.global.backends do 
local s = proxy.global.backends[i] 
local pool = s.pool -- we don't have a username yet, try to find a connections which is idling 
local cur_idle = pool.users[""].cur_idle_connections 

pool.min_idle_connections = proxy.global.config.rwsplit.min_idle_connections 
pool.max_idle_connections = proxy.global.config.rwsplit.max_idle_connections 


if is_debug then 
print(" [".. i .."].connected_clients = " .. s.connected_clients) 
print(" [".. i .."].pool.cur_idle = " .. cur_idle) 
print(" [".. i .."].pool.max_idle = " .. pool.max_idle_connections) 
print(" [".. i .."].pool.min_idle = " .. pool.min_idle_connections) 
print(" [".. i .."].type = " .. s.type) 
print(" [".. i .."].state = " .. s.state) 
end 
-- prefer connections to the master 
if s.type == proxy.BACKEND_TYPE_RW and 
s.state ~= proxy.BACKEND_STATE_DOWN and 
cur_idle < pool.min_idle_connections then 
proxy.connection.backend_ndx = i 
break 
elseif s.type == proxy.BACKEND_TYPE_RO and 
s.state ~= proxy.BACKEND_STATE_DOWN and 
cur_idle < pool.min_idle_connections then 
proxy.connection.backend_ndx = i 
break 
elseif s.type == proxy.BACKEND_TYPE_RW and 
s.state ~= proxy.BACKEND_STATE_DOWN and 
rw_ndx == 0 then 
rw_ndx = i 
end 
end 


if proxy.connection.backend_ndx == 0 then 
if is_debug then 
print(" [" .. rw_ndx .. "] taking master as default") 
end 
proxy.connection.backend_ndx = rw_ndx 
end 


-- pick a random backend 
-- 
-- we someone have to skip DOWN backends 


-- ok, did we got a backend ? 


if proxy.connection.server then 
if is_debug then 
print(" using pooled connection from: " .. proxy.connection.backend_ndx) 
end 


-- stay with it 
return proxy.PROXY_IGNORE_RESULT 
end 


if is_debug then 
print(" [" .. proxy.connection.backend_ndx .. "] idle-conns below min-idle") 
end 


-- open a new connection 
end 


--- 
-- put the successfully authed connection into the connection pool 
-- 
-- @param auth the context information for the auth 
-- 
-- auth.packet is the packet 
function read_auth_result( auth ) 
if is_debug then 
print("[read_auth_result] " .. proxy.connection.client.src.name) 
end 
if auth.packet:byte() == proxy.MYSQLD_PACKET_OK then 
-- auth was fine, disconnect from the server 
proxy.connection.backend_ndx = 0 
elseif auth.packet:byte() == proxy.MYSQLD_PACKET_EOF then 
-- we received either a 
-- 
-- * MYSQLD_PACKET_ERR and the auth failed or 
-- * MYSQLD_PACKET_EOF which means a OLD PASSWORD (4.0) was sent 
print("(read_auth_result) ... not ok yet"); 
elseif auth.packet:byte() == proxy.MYSQLD_PACKET_ERR then 
-- auth failed 
end 
end 
--- 
-- read/write splitting 
function read_query( packet ) 
local is_debug = proxy.global.config.rwsplit.is_debug 
local cmd = commands.parse(packet) 
local c = proxy.connection.client 


local r = auto_config.handle(cmd) 
if r then return r end 


local tokens 
local norm_query 


-- looks like we have to forward this statement to a backend 
if is_debug then 
print("[read_query] " .. proxy.connection.client.src.name) 
print(" current backend = " .. proxy.connection.backend_ndx) 
print(" client default db = " .. c.default_db) 
print(" client username = " .. c.username) 
if cmd.type == proxy.COM_QUERY then 
print(" query = " .. cmd.query) 
end 
end 


if cmd.type == proxy.COM_QUIT then 
-- don't send COM_QUIT to the backend. We manage the connection 
-- in all aspects. 
proxy.response = { 
type = proxy.MYSQLD_PACKET_OK, 
} 


if is_debug then 
print(" (QUIT) current backend = " .. proxy.connection.backend_ndx) 
end 


return proxy.PROXY_SEND_RESULT 
end 


-- COM_BINLOG_DUMP packet can't be balanced 
-- 
-- so we must send it always to the master 
if cmd.type == proxy.COM_BINLOG_DUMP then 
-- if we don't have a backend selected, let's pick the master 
-- 
if proxy.connection.backend_ndx == 0 then 
proxy.connection.backend_ndx = lb.idle_failsafe_rw() 
end 


return 
end 


proxy.queries:append(1, packet, { resultset_is_needed = true }) 


-- read/write splitting 
-- 
-- send all non-transactional SELECTs to a slave 
if not is_in_transaction and 
cmd.type == proxy.COM_QUERY then 
tokens = tokens or assert(tokenizer.tokenize(cmd.query)) 


local stmt = tokenizer.first_stmt_token(tokens) 


if stmt.token_name == "TK_SQL_SELECT" then 
is_in_select_calc_found_rows = false 
local is_insert_id = false 


for i = 1, #tokens do 
local token = tokens[i] 
-- SQL_CALC_FOUND_ROWS + FOUND_ROWS() have to be executed 
-- on the same connection 
-- print("token: " .. token.token_name) 
-- print(" val: " .. token.text) 


if not is_in_select_calc_found_rows and token.token_name == "TK_SQL_SQL_CALC_FOUND_ROWS" then 
is_in_select_calc_found_rows = true 
elseif not is_insert_id and token.token_name == "TK_LITERAL" then 
local utext = token.text:upper() 


if utext == "LAST_INSERT_ID" or 
utext == "@@INSERT_ID" then 
is_insert_id = true 
end 
end 


-- we found the two special token, we can't find more 
if is_insert_id and is_in_select_calc_found_rows then 
break 
end 
end 


-- if we ask for the last-insert-id we have to ask it on the original 
-- connection 
if not is_insert_id then 
local backend_ndx = lb.idle_ro() 


if backend_ndx > 0 then 
proxy.connection.backend_ndx = backend_ndx 
end 
else 
print(" found a SELECT LAST_INSERT_ID(), staying on the same backend") 
end 
end 
end 


-- no backend selected yet, pick a master 
if proxy.connection.backend_ndx == 0 then 
-- we don't have a backend right now 
-- 
-- let's pick a master as a good default 
-- 
proxy.connection.backend_ndx = lb.idle_failsafe_rw() 
end 


-- by now we should have a backend 
-- 
-- in case the master is down, we have to close the client connections 
-- otherwise we can go on 
if proxy.connection.backend_ndx == 0 then 
return proxy.PROXY_SEND_QUERY 
end 


local s = proxy.connection.server 


-- if client and server db don't match, adjust the server-side 
-- 
-- skip it if we send a INIT_DB anyway 
if cmd.type ~= proxy.COM_INIT_DB and 
c.default_db and c.default_db ~= s.default_db then 
print(" server default db: " .. s.default_db) 
print(" client default db: " .. c.default_db) 
print(" syncronizing") 
proxy.queries:prepend(2, string.char(proxy.COM_INIT_DB) .. c.default_db, { resultset_is_needed = true }) 
end 


-- send to master 
if is_debug then 
if proxy.connection.backend_ndx > 0 then 
local b = proxy.global.backends[proxy.connection.backend_ndx] 
print(" sending to backend : " .. b.dst.name); 
print(" is_slave : " .. tostring(b.type == proxy.BACKEND_TYPE_RO)); 
print(" server default db: " .. s.default_db) 
print(" server username : " .. s.username) 
end 
print(" in_trans : " .. tostring(is_in_transaction)) 
print(" in_calc_found : " .. tostring(is_in_select_calc_found_rows)) 
print(" COM_QUERY : " .. tostring(cmd.type == proxy.COM_QUERY)) 
end 


return proxy.PROXY_SEND_QUERY 
end 


--- 
-- as long as we are in a transaction keep the connection 
-- otherwise release it so another client can use it 
function read_query_result( inj ) 
local is_debug = proxy.global.config.rwsplit.is_debug 
local res = assert(inj.resultset) 
local flags = res.flags 


if inj.id ~= 1 then 
-- ignore the result of the USE <default_db> 
-- the DB might not exist on the backend, what do do ? 
-- 
if inj.id == 2 then 
-- the injected INIT_DB failed as the slave doesn't have this DB 
-- or doesn't have permissions to read from it 
if res.query_status == proxy.MYSQLD_PACKET_ERR then 
proxy.queries:reset() 


proxy.response = { 
type = proxy.MYSQLD_PACKET_ERR, 
errmsg = "can't change DB ".. proxy.connection.client.default_db .. 
" to on slave " .. proxy.global.backends[proxy.connection.backend_ndx].dst.name 
} 


return proxy.PROXY_SEND_RESULT 
end 
end 
return proxy.PROXY_IGNORE_RESULT 
end 


is_in_transaction = flags.in_trans 
local have_last_insert_id = (res.insert_id and (res.insert_id > 0)) 


if not is_in_transaction and 
not is_in_select_calc_found_rows and 
not have_last_insert_id then 
-- release the backend 
proxy.connection.backend_ndx = 0 
elseif is_debug then 
print("(read_query_result) staying on the same backend") 
print(" in_trans : " .. tostring(is_in_transaction)) 
print(" in_calc_found : " .. tostring(is_in_select_calc_found_rows)) 
print(" have_insert_id : " .. tostring(have_last_insert_id)) 
end 
end 


--- 
-- close the connections if we have enough connections in the pool 
-- 
-- @return nil - close connection 
-- IGNORE_RESULT - store connection in the pool 
function disconnect_client() 
local is_debug = proxy.global.config.rwsplit.is_debug 
if is_debug then 
print("[disconnect_client] " .. proxy.connection.client.src.name) 
end 


-- make sure we are disconnection from the connection 
-- to move the connection into the pool 
proxy.connection.backend_ndx = 0 
end