mysql4.0做主从时主库的备份脚本_MySQL
程序员文章站
2024-02-13 13:47:28
...
bitsCN.com
mysql4.0做主从时主库的备份脚本 mysql4.0是老版本了，但是有些早期使用的企业依然在用，在创建主从时特别是线上服务器创建主从时，保证数据的一致性是个大问题：比如创建完从库同步时出现重复数据重复执行（虽然数据条数一致，但数据有可能会不一致）等。 在mysql5.0以上版本中，此时备份主库只用在mysqldump时加上-F、master-data=2，single-transaction参数，从库同步时导入备份，在取备份文件开头的bin-log和pos位置进行同步即可，不会出现数据重复执行等问题，他能确保同步时的一致性。比较悲剧的是，本人所用的数据库还没有升级，是4.0的版本，经过测试，写了一个专一用于4.0主从同步时主库备份的脚本，原理就是模拟5.0以上的备份过程来做的。也可以用于5.0以上的版本，但是5.0以上的版本没有必要这么做。大家可以参考。 #!/usr/bin/env python # -*- coding: utf-8 -*- import os,sys,time,MySQLdb import subprocess,threading class mysql_dump(): def __init__(self): self.dumpname = "/root/alldata%s.bz2" % time.strftime("%Y%m%d%H%M") self.STAT_IP = "192.168.0.39" self.logfile = "/root/mysql_dump.log" self.user = "root" self.passwd = "1q2w3e4r" def log_w(self,text): now = time.strftime("%Y-%m-%d %H:%M:%S") tt = str(now) + "/t" + text + "/n" f = open(self.logfile,'a+') f.write(tt) f.close() def dump(self): cmd = "/usr/local/mysql/bin/mysqldump -A -Q -e --add-drop-table --add-locks --extended-insert --quick --no-autocommit --single-transaction -u%s -p%s | bzip2 -2 > %s" % (self.user,self.passwd,self.dumpname) print time.strftime("%Y-%m-%d %H:%M:%S") text = "Start mysqldump,Please wait ..." print text self.log_w(text) a = subprocess.Popen(cmd,shell=True) while 1: b = subprocess.Popen.poll(a) if b == 0: text = "Mysqldump complete" print text self.log_w(text) break elif b is None: print 'Mysqldump running' time.sleep(30) else: print a.pid,'term' break self.rsync() def rsync(self): cmd = "rsync -az %s %s::asktao_db/db_back/" % (self.dumpname,self.STAT_IP) text = "Start rsync to server(%s) ,Please wait ..." % self.STAT_IP print text self.log_w(text) a = subprocess.Popen(cmd,shell=True) while 1: b = subprocess.Popen.poll(a) if b == 0: text = "Rsync complete" print text self.log_w(text) break elif b is None: print 'Rsync running' time.sleep(30) else: print a.pid,'term' break def lock(self): try: conn = MySQLdb.connect(host = '127.0.0.1',user = 'root',passwd = '1q2w3e4r', charset='utf8', connect_timeout=5) cursor = conn.cursor() text = "flush tables with read lock" print text self.log_w(text) cursor.execute("flush tables with read lock") text = "flush logs" print text self.log_w(text) cursor.execute("flush logs") d = threading.Thread(target=self.dump, args=()) d.start() while 1: if os.path.isfile(self.dumpname) and os.path.getsize(self.dumpname) > 0: text = "UNLOCK TABLES" print text self.log_w(text) cursor.execute("UNLOCK TABLES") break except MySQLdb.Error,e: text = e.args print text self.log_w(text) def work(self): t = threading.Thread(target=self.lock, args=()) t.start() if __name__ == "__main__": boss = mysql_dump() boss.work() bitsCN.com
mysql4.0做主从时主库的备份脚本 mysql4.0是老版本了，但是有些早期使用的企业依然在用，在创建主从时特别是线上服务器创建主从时，保证数据的一致性是个大问题：比如创建完从库同步时出现重复数据重复执行（虽然数据条数一致，但数据有可能会不一致）等。 在mysql5.0以上版本中，此时备份主库只用在mysqldump时加上-F、master-data=2，single-transaction参数，从库同步时导入备份，在取备份文件开头的bin-log和pos位置进行同步即可，不会出现数据重复执行等问题，他能确保同步时的一致性。比较悲剧的是，本人所用的数据库还没有升级，是4.0的版本，经过测试，写了一个专一用于4.0主从同步时主库备份的脚本，原理就是模拟5.0以上的备份过程来做的。也可以用于5.0以上的版本，但是5.0以上的版本没有必要这么做。大家可以参考。 #!/usr/bin/env python # -*- coding: utf-8 -*- import os,sys,time,MySQLdb import subprocess,threading class mysql_dump(): def __init__(self): self.dumpname = "/root/alldata%s.bz2" % time.strftime("%Y%m%d%H%M") self.STAT_IP = "192.168.0.39" self.logfile = "/root/mysql_dump.log" self.user = "root" self.passwd = "1q2w3e4r" def log_w(self,text): now = time.strftime("%Y-%m-%d %H:%M:%S") tt = str(now) + "/t" + text + "/n" f = open(self.logfile,'a+') f.write(tt) f.close() def dump(self): cmd = "/usr/local/mysql/bin/mysqldump -A -Q -e --add-drop-table --add-locks --extended-insert --quick --no-autocommit --single-transaction -u%s -p%s | bzip2 -2 > %s" % (self.user,self.passwd,self.dumpname) print time.strftime("%Y-%m-%d %H:%M:%S") text = "Start mysqldump,Please wait ..." print text self.log_w(text) a = subprocess.Popen(cmd,shell=True) while 1: b = subprocess.Popen.poll(a) if b == 0: text = "Mysqldump complete" print text self.log_w(text) break elif b is None: print 'Mysqldump running' time.sleep(30) else: print a.pid,'term' break self.rsync() def rsync(self): cmd = "rsync -az %s %s::asktao_db/db_back/" % (self.dumpname,self.STAT_IP) text = "Start rsync to server(%s) ,Please wait ..." % self.STAT_IP print text self.log_w(text) a = subprocess.Popen(cmd,shell=True) while 1: b = subprocess.Popen.poll(a) if b == 0: text = "Rsync complete" print text self.log_w(text) break elif b is None: print 'Rsync running' time.sleep(30) else: print a.pid,'term' break def lock(self): try: conn = MySQLdb.connect(host = '127.0.0.1',user = 'root',passwd = '1q2w3e4r', charset='utf8', connect_timeout=5) cursor = conn.cursor() text = "flush tables with read lock" print text self.log_w(text) cursor.execute("flush tables with read lock") text = "flush logs" print text self.log_w(text) cursor.execute("flush logs") d = threading.Thread(target=self.dump, args=()) d.start() while 1: if os.path.isfile(self.dumpname) and os.path.getsize(self.dumpname) > 0: text = "UNLOCK TABLES" print text self.log_w(text) cursor.execute("UNLOCK TABLES") break except MySQLdb.Error,e: text = e.args print text self.log_w(text) def work(self): t = threading.Thread(target=self.lock, args=()) t.start() if __name__ == "__main__": boss = mysql_dump() boss.work() bitsCN.com