MySQL数据库主从切换脚本自动化
在一些实际环境中，如何实现主从的快速切换，在没有MHA等工具的辅助下，如何避免影响线上的业务，一般都会在在业务低峰期进行主从切换，本脚本主要利用MySQL自带的命令行工具(FLUSH TABLES WITH READ LOCK)进行锁全库，且由用户自行输入判断多少秒内从库BINLOG数据不在同步后，认为主从数据已达一致性可以进行主从切换(在一些资料上说也可以用READ-ONLY来锁库，但需要注意如果写入用户具有Admin权限是不受限制),主从切换完成5秒后进行检查主从是否同步问题.
##连接数据库使用TCP/IP方式，需要相应的操作数据库权限。
##目前判断使用show master status 来判断，可以show slave status返回的值来判断，自行修改。
脚本:
#!/usr/bin/env python # -*- coding: utf-8 -*- import MySQLdb import time import sys class m_s: def __init__(self,host,user,password,port): self.host=host self.user=user self.passowrd=password self.port=port def getConn(self,db="mysql"): try: conn=MySQLdb.connect(host=self.host, user=self.user, passwd=self.passowrd, db=db, port=self.port, charset="utf8") cur = conn.cursor() return cur except Exception as e: return e def execSQLlock(self,*args): flush_sql="FLUSH TABLES WITH READ LOCK" cur.execute(flush_sql) def execIo(self,cur,command): cur.execute(command) db_pos = cur.fetchall() for value in db_pos: value=value return value def exeStop(self,cur,command): cur.execute(command) db_pos = cur.fetchall() return db_pos def execSQLstatus(self,*args): n=0 self.execSQLlock(cur) flush_m = "flush logs" cur.execute(flush_m) while True: data=[] slave_pos=[] n=n+1 exe_sql = "select Command,State,Info,Id from information_schema.processlist" cur.execute(exe_sql) plist = cur.fetchall() for li in range(len(plist)): if plist[li][0] == "Query" and plist[li][1] == "Waiting for global read lock": lock_id = "kill " + str(plist[li][3]) print plist[li][2] cur.execute(lock_id) slave_pos.append(self.execIo(cur1, "show master status")[1]) data.append(self.execIo(cur1,"show slave status")[6]) ##从库的游标 time.sleep(1) slave_pos.append(self.execIo(cur1, "show master status")[1]) data.append(self.execIo(cur,"show master status")[1])##从库的游标 print ".......",data,slave_pos if data[0]==data[1] and slave_pos[0]==slave_pos[1]: try: print "第%s次判断数据已经同步....."%n if n==c_time: print "开始主从切换工作........" self.exeStop(cur1,"stop slave") ##停止从库同步 new_pos=self.exeStop(cur1,"show master status")#获取新主库的FILE和POS的值,游标为还没切换前的从库 #print "获取新主库的FILE和POS的值,游标为还没切换前的从库",new_pos self.exeStop(cur,"reset master;")##主库释放从库主从信息...... ##在从库执行new_change 指向新的主库 self.exeStop(cur1,"reset slave all") new_change="change master to master_host='"+str(args[1])+"'"+",master_user='"+args[2]+"'"+",master_password='"+args[3]+"',master_port="+str(args[4])+",MASTER_LOG_FILE='"+str(new_pos[0][0])+"'"+",MASTER_LOG_POS="+str(new_pos[0][1]) print new_change self.exeStop(cur,new_change)##在原来主库上执行change master to..... #print "在原来主库上执行change master to...." self.exeStop(cur,"start slave") ##在原来主库上执行change master to..... time.sleep(5) s_pos=self.exeStop(cur,"show slave status;") #print s_pos[0][10],s_pos[0][11] if s_pos[0][10]=="Yes" and s_pos[0][11]=="Yes": self.exeStop(cur1,"reset slave all") print "主从切换成功!" print "\n" while True: print "等待其他操作完成,即将unlock tables主库......" try: stop = raw_input("输入终止命令q即完成此次操作:\n") if stop == "q": sys.exit() # break except Exception as e: print "good bye" else: print s_pos[0][19] break except Exception as e: return e else: print "主从数据未达到一致性..........",n n=0 data=[] if __name__ =="__main__": c_time=int(raw_input("多少秒后进行主从切换..>>")) print "****************************************************\n" print "请根据提示输入指定信息:" m_host = raw_input("目前主库的地址:") m_user = raw_input("目前主库的登陆用户名:") m_password = raw_input("目前主库的密码:") m_port = int(raw_input("目前主库的端口:")) print "****************************************************\n" s_host = raw_input("目前从库的地址:") s_user = raw_input("目前从库的登陆用户名:") s_password = raw_input("目前从库的密码:") s_port = int(raw_input("目前从库的端口:")) M = m_s(m_host, m_user, m_password, m_port) S = m_s(s_host,s_user,s_password,s_port) cur = M.getConn() ##获取主库游标 cur1 = S.getConn() ##获取从库游标 print "*****************************************************\n" print "主从同步用户信息.........\n" s_user1 = raw_input("输入主从同步的用户>>:") s_password1 = raw_input("输入主从同步的密码:") s_port1 = int(raw_input("输入主从同步的端口:")) M.execSQLstatus(cur,s_host,s_user1,s_password1,s_port1)