使用mysql-proxy 快速实现mysql 集群 读写分离

关于mysql、mysql主从的搭建，在此不再演示，如下的操作均在mysql-proxy（192.168.1.200）服务器进行 
一、安装mysql-proxy 
1、安装lua  (mysql-proxy需要使用lua脚本进行数据转发) 
#tar zxvf lua-5.1.4.tar.gz 
#cd lua-5.1.4 
#vi Makefile，修改INSTALL_TOP= /usr/local/lua 
#make posix 
#make install 
 
2、安装libevent 
#tar zxvf libevent-2.0.8-rc.tar.gz 
#cd libevent-2.0.8-rc 
#./configure --prefix=/usr/local/libevent 
#make && make install 
 
3、安装check 
#tar zxvf check-0.9.8.tar.gz 
#cd check-0.9.8 
#./configure && make && make install 
 
4、安装mysql客户端 
#tar zxvf mysql-5.0.92.tar.gz 
#cd mysql-5.0.92 
#./configure --without-server && make && make install 
 
5、设置环境变量 （安装mysql-proxy所需变量） 
#vi /etc/profile 
export LUA_CFLAGS="-I/usr/local/lua/include" LUA_LIBS="-L/usr/local/lua/lib -llua -ldl" LDFLAGS="-L/usr/local/libevent/lib -lm" 
export CPPFLAGS="-I/usr/local/libevent/include" 
export CFLAGS="-I/usr/local/libevent/include" 
# source /etc/profile 
 
6、安装mysql-proxy 
#tar zxvf mysql-proxy-0.6.0.tar.gz 
#cd mysql-proxy-0.6.0 
# ./configure --prefix=/usr/local/mysql-proxy --with-mysql --with-lua 
#make && make install 
 
7、启动mysql-proxy 
本次对两台数据库实现了读写分离；mysql-master为可读可写，mysql-slave为只读 
#/usr/local/mysql-proxy/sbin/mysql-proxy --proxy-backend-addresses=192.168.1.201:3306 --proxy-read-only-backend-addresses=192.168.1.202:3306 --proxy-lua-script=/usr/local/mysql-proxy/share/mysql-proxy/rw-splitting.lua &  
 
注：如果正常情况下启动后终端不会有任何提示信息，mysql-proxy启动后会启动两个端口4040和4041，4040用于SQL转发，4041用于管理mysql-proxy。如有多个mysql-slave可以依次在后面添加 
 
 
二、测试 
1、连接测试 
因为默认情况下mysql数据库不允许用户在远程连接 
mysql>grant all privileges on *.* to identified by '123456'; 
mysql>flush privileges; 
 
客户端连接 
#mysql -uroot -p123456 -h192.168.1.200 -P4040 
 
 
2、读写分离测试 
为了测试出mysql读写分离的真实性，在测试之前，需要开启两台mysql的log功能，然后在mysql-slave服务器停止复制 
① 、在两台mysql配置文件my.cnf中加入log=query.log，然后重启

② 、在mysql-slave上执行SQL语句stop slave

③ 、在两台mysql上执行#tail -f /usr/local/mysql/var/query.log

④ 、在客户端上连接mysql（三个连接以上），然后执行create、select等SQL语句，观察两台mysql的日志有何变化

注：生产环境中除了进行程序调试外，其它不要开启mysql查询日志，因为查询日志记录了客户端的所有语句，频繁的IO操作将会导致mysql整体性能下降 
 
总结：在上述环境中，mysql-proxy和mysql-master、mysql-slave三台服务器均存在单点故障。如果在可用性要求较高的场合，单点隐患是绝对不允许的。为了避免mysql-proxy单点隐患有两种方法，一种方法是mysql-proxy配合keepalived做双机，另一种方法是将mysql-proxy和应用服务安装到同一台服务器上；为了避免mysql-master单点故障可以使用DRBD+heartbear做双机；避免mysql-slave单点故障增加多台mysql-slave即可，因为mysql-proxy会自动屏蔽后端发生故障的mysql-slave。

 

附: mysql-proxy LUA 读写分离脚本代码:

--[[
--
-- author : KDr2 
-- version 0.01
-- SYNOPSIS:
---  1.维护了一个连接池
---  2.读写分离，简单的将select开头的语句放到slave上执行
---  3.事务支持，所有事务放到master上执行，事务中不更改连接
---  4.简单日志
--
--]]

--- config vars
local min_idle_connections = 4
local max_idle_connections = 8
local log_level=1
local encoding="utf8"
--- end of config

-- 事务标识，在事务内不归还连接
local transaction_flags={}
setmetatable(transaction_flags,{__index=function() return 0 end})

-- log system
log={
   level={debug=1,info=2,warn=3,error=4},
   funcs={"debug","info","warn","error"},
}
function log.log(level,m)
   if level >= log_level then
      local msg="[" .. os.date("%Y-%m-%d %X") .."] ".. log.funcs[level] .. ": " .. tostring(m)
      print(msg) -- TODO  write msg into a log file.
   end
end
for i,v in ipairs(log.funcs) do
   log[v]=function(m) log.log(log.level[v],m) end
end

-- connect to server
function connect_server() 
   log.info(" starting connect_server ... ")
   local least_idle_conns_ndx = 0
   local least_idle_conns = 0
   
   for i = 1, #proxy.backends do
      local s = proxy.backends[i]
      local pool = s.pool 
      local cur_idle = pool.users[""].cur_idle_connections

      log.debug("[".. s.address .."].connected_clients = " .. s.connected_clients)
      log.debug("[".. s.address .."].idling_connections = " .. cur_idle)
      log.debug("[".. s.address .."].type = " .. s.type)
      log.debug("[".. s.address .."].state = " .. s.state)

      if s.state ~= proxy.BACKEND_STATE_DOWN then
         -- try to connect to each backend once at least
         if cur_idle == 0 then
            proxy.connection.backend_ndx = i
            log.info("server [".. proxy.backends[i].address .."] open new connection")
            return
         end
         -- try to open at least min_idle_connections
         if least_idle_conns_ndx == 0 or
            ( cur_idle < min_idle_connections and 
              cur_idle < least_idle_conns ) then
            least_idle_conns_ndx = i
            least_idle_conns = cur_idle
         end
      end
   end

   if least_idle_conns_ndx > 0 then
      proxy.connection.backend_ndx = least_idle_conns_ndx
   end
   
   if proxy.connection.backend_ndx > 0 then 
      local s = proxy.backends[proxy.connection.backend_ndx]
      local pool = s.pool 
      local cur_idle = pool.users[""].cur_idle_connections

      if cur_idle >= min_idle_connections then
         -- we have 4 idling connections in the pool, that's good enough
         log.debug("using pooled connection from: " .. proxy.connection.backend_ndx)
         return proxy.PROXY_IGNORE_RESULT
      end
   end
   -- open a new connection 
   log.info("opening new connection on: " .. proxy.backends[proxy.connection.backend_ndx].address)
end

---

-- auth.packet is the packet
function read_auth_result( auth )
   if auth.packet:byte() == proxy.MYSQLD_PACKET_OK then
      -- 连接正常
      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
      log.error("(read_auth_result) ... not ok yet");
   elseif auth.packet:byte() == proxy.MYSQLD_PACKET_ERR then
      log.error("auth failed!")
   end
end

--- 
-- read/write splitting
function read_query( packet ) 
   log.debug("[read_query]")
   log.debug("authed backend = " .. proxy.connection.backend_ndx)
   log.debug("used db = " .. proxy.connection.client.default_db)

   if packet:byte() == proxy.COM_QUIT then
      proxy.response = {
         type = proxy.MYSQLD_PACKET_OK,
      }
      return proxy.PROXY_SEND_RESULT
   end

   if proxy.connection.backend_ndx == 0 then
      local is_read=(string.upper(packet:sub(2))):match("^SELECT")
      local target_type=proxy.BACKEND_TYPE_RW
      if is_read then target_type=proxy.BACKEND_TYPE_RO end
      for i = 1, #proxy.backends do
         local s = proxy.backends[i]
         local pool = s.pool 
         local cur_idle = pool.users[proxy.connection.client.username].cur_idle_connections
         
         if cur_idle > 0 and 
            s.state ~= proxy.BACKEND_STATE_DOWN and 
            s.type == target_type then
            proxy.connection.backend_ndx = i
            break
         end
      end
   end
   -- sync the client-side default_db with the server-side default_db
   if proxy.connection.server and proxy.connection.client.default_db ~= proxy.connection.server.default_db then
      local server_db=proxy.connection.server.default_db
      local client_db=proxy.connection.client.default_db
      local default_db= (#client_db > 0) and client_db or server_db
      if #default_db > 0 then
         proxy.queries:append(2, string.char(proxy.COM_INIT_DB) .. default_db)
         proxy.queries:append(2, string.char(proxy.COM_QUERY) .. "set names '" .. encoding .."'")
         log.info("change database to " .. default_db);
      end
   end
   if proxy.connection.backend_ndx > 0 then
      log.debug("Query[" .. packet:sub(2) .. "] Target is [" .. proxy.backends[proxy.connection.backend_ndx].address .."]")
   end
   proxy.queries:append(1, packet)
   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 res      = assert(inj.resultset)
   local flags    = res.flags

   if inj.id ~= 1 then
      -- ignore the result of the USE <default_db>
      return proxy.PROXY_IGNORE_RESULT
   end
   is_in_transaction = flags.in_trans

   if flags.in_trans then
      transaction_flags[proxy.connection.server.thread_id] = transaction_flags[proxy.connection.server.thread_id] + 1
   elseif inj.query:sub(2):lower():match("^%s*commit%s*$") or inj.query:sub(2):lower():match("^%s*rollback%s*$") then
      transaction_flags[proxy.connection.server.thread_id] = transaction_flags[proxy.connection.server.thread_id] - 1
      if transaction_flags[proxy.connection.server.thread_id] < 0 then transaction_flags[proxy.connection.server.thread_id] = 0 end
   end
   
   log.debug("transaction res : " .. tostring(transaction_flags[proxy.connection.server.thread_id]));
   if transaction_flags[proxy.connection.server.thread_id]==0 or transaction_flags[proxy.connection.server.thread_id] == nil then 
      -- isnot in a transaction, need to release the backend
      proxy.connection.backend_ndx = 0
   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()
   log.debug("[disconnect_client]")
   if proxy.connection.backend_ndx == 0 then
      for i = 1, #proxy.backends do
         local s = proxy.backends[i]
         local pool = s.pool 
         local cur_idle = pool.users[proxy.connection.client.username].cur_idle_connections
         
         if s.state ~= proxy.BACKEND_STATE_DOWN and
            cur_idle > max_idle_connections then
            -- try to disconnect a backend
            proxy.connection.backend_ndx = i
            log.info("[".. proxy.backends[i].address .."] closing connection, idling: " .. cur_idle)
            return
         end
      end
      return proxy.PROXY_IGNORE_RESULT
   end
end