缘由:
测试环境中,主库执行了DDL语句增加一个字段的长度后,从库报无法修改这个字段长的的问题,且这个字段的长度已经介于原来的长度和目标长度中间了
环境
5.7.19 GTID双主复制
解决步骤:
1.尝试手工修改字段长度,恢复到未修改前的长度,重启slave进程。结果:失败,报同样的错误,错误编号1677
2.尝试手工修改字段长度,同步到修改后长度,跳过这个事务:
mysql>stop slave; mysql>set gtid_next="d7c35015-9dd1-11e7-b70d-005056aa19c3:51629"; ##这里需要注意,由于开启了双主GTID复制,show master status和Executed_gtid_set会有两个GTID值,其中一个为自己的GTID,另一个为主的GTID。 ##设置的时候只要把主库的GTID写进“”即可。自身的GTID不需要指明,但如果使用set gtid_purged的方式跳过,是需要可以指明两个GTID的 mysql>begin;commit; ##设置后,插入一个空事务进行更新GTID。 mysql>set gtid_next=‘automatic‘; ##官方手册规定:精确的指定过一次GTID,并产生一次事务后,后面必须再次指定一次gtid_next的模式(是模式,不是精确值,官方手册这里没有写清楚) ##“After this variable has been set to UUID:NUMBER, and a transaction has been committed or rolled back, an explicit SET GTID_NEXT statement must again be issued before any other statement.” mysql>start slave;
讲道理,这里就应该正常了,但还是报同样的错误,一次次的这样操作也很麻烦,
3.批量跳过复制错误有如下两个方法
使用slave-skip-errors=123,456,789,但这个参数不是动态参数,需要写进配置文件并重启,而且也不方便观测
使用percona公司的pt-slave-restart工具
pt-slave-restart是percona-toolkit工具集中的一个专用于处理复制错误的工具
原理:根据设置,跳过从主库过来的指定错误事务
1.支持GTID复制,但是不支持多线程复制,工具分不清到底哪个线程复制出了问题
2.除了主机指定外的实用参数:
--always :永不停止slave线程,手工停止也不行
--ask-pass :替换-p命令,防止密码输入被身后的开发窥屏
--error-numbers :指定跳过哪些错误,可用,进行分隔
--error-text :根据错误信息进行匹配跳过
--log :输出到文件
--recurse :在主端执行,监控从端
--runtime :工具执行多长时间后退出:默认秒, m=minute,h=hours,d=days
--slave-user --slave-password :从库的账号密码,从主端运行时使用
--skip-count :一次跳过错误的个数,胆大的可以设置大些,不指定默认1个
--master-uuid :级联复制的时候,指定跳过上级或者上上级事务的错误
--until-master :到达指定的master_log_pos,file位置后停止,
格式:”file:pos“
--until-relay :和上面一样,但是时根据relay_log的位置来停止
安装:
centos#yum install https://www.percona.com/redir/downloads/percona-release/redhat/percona-release-0.1-4.noarch.rpm centos#yum search percon-tool centos#yum -y install yum install percona-toolkit
查看帮助文件:
perldoc /usr/bin/pt-slave-restart
运行工具:
pt-slave-restart --user=root --password=123456 --socket=/data/mysql/3304/tmp/mysql3304.sock --error-numbers=1677
输出如下:
Not checking slave because relay log file or position has not changed (file relay-bin.000002 pos 205804892) 2017-09-30T12:55:03 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 205805634 1677 Not checking slave because relay log file or position has not changed (file relay-bin.000002 pos 205805634) Not checking slave because relay log file or position has not changed (file relay-bin.000002 pos 205805634) Not checking slave because relay log file or position has not changed (file relay-bin.000002 pos 205805634) 2017-09-30T12:55:03 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 205806382 1677 2017-09-30T12:55:04 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 205807125 1677 2017-09-30T12:55:04 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 205807873 1677 Not checking slave because relay log file or position has not changed (file relay-bin.000002 pos 205807873) 2017-09-30T12:55:20 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 205997621 1677 2017-09-30T12:55:21 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 205998353 1677 2017-09-30T12:55:22 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 205999085 1677 2017-09-30T12:55:23 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 205999823 1677 2017-09-30T12:55:24 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 206000565 1677 2017-09-30T12:55:24 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 206001307 1677 2017-09-30T12:55:24 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 206002049 1677 2017-09-30T12:55:24 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 206002783 1677 2017-09-30T12:55:24 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 206003517 1677 2017-09-30T12:55:24 S=/data/mysql/3304/tmp/mysql3304.sock,p=...,u=root relay-bin.000002 206004253 1677
查看从库线程情况:
************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 192.168.100.200 Master_User: bt_repl Master_Port: 3304 Connect_Retry: 60 Master_Log_File: 3304binlog.000002 Read_Master_Log_Pos: 665635124 Relay_Log_File: relay-bin.000006 Relay_Log_Pos: 2080902 Relay_Master_Log_File: 3304binlog.000002 Slave_IO_Running: Yes Slave_SQL_Running: Yes Last_Errno: 0 Last_Error: Skip_Counter: 0 Exec_Master_Log_Pos: 665635124 Relay_Log_Space: 2081454 Until_Condition: None Until_Log_File: Until_Log_Pos: 0 Master_SSL_Allowed: No Seconds_Behind_Master: 0 Master_SSL_Verify_Server_Cert: No Last_IO_Errno: 0 Last_IO_Error: Last_SQL_Errno: 0 Last_SQL_Error: Replicate_Ignore_Server_Ids: Master_Server_Id: 3304 Master_UUID: d7c35015-9dd1-11e7-b70d-005056aa19c3 Master_Info_File: mysql.slave_master_info SQL_Delay: 0 SQL_Remaining_Delay: NULL Slave_SQL_Running_State: Slave has read all relay log; waiting for more updates Master_Retry_Count: 86400 Master_Bind: Retrieved_Gtid_Set: d7c35015-9dd1-11e7-b70d-005056aa19c3:1-430778 Executed_Gtid_Set: 886f86e9-9dd6-11e7-ba30-005056aa6236:1-2, d7c35015-9dd1-11e7-b70d-005056aa19c3:1-430778 Auto_Position: 1 1 row in set (0.07 sec)
已经没有错误了。OK
attention:
1.gtid_next是一个会话级别的参数,而gtid_purged则是一个全局级别的参数
本文出自 “漫漫SQL路......” 博客,请务必保留此出处http://l0vesql.blog.51cto.com/4159433/1970042
一次主从复制出错解决与pt-slave-restart工具使用
原文地址:http://l0vesql.blog.51cto.com/4159433/1970042