mysql-mmm配置文档--下

四、主主（master-master）同步配置

1）172.16.71.132机器my.cnf

skip-name-resolve

server_id           = 132

set-variable = key_buffer_size=512M

set-variable = table_cache=32

set-variable = sort_buffer_size=32M

set-variable = read_buffer_size=32M

set-variable = max_connections=2100

set-variable = log_slow_queries=/var/lib/mysql/log/mysql-slow.log

set-variable = log_bin=/var/lib/mysql/log/mysql-bin

set-variable = long_query_time=1

set-variable = thread_concurrency=16

set-variable = tmp_table_size=4000000000

set-variable = slave_skip_errors=all

#replicate-ignore-db=mysql

#replicate-ignore-db=test  发觉加上这个ignore选项后会导致mmm接管出现点小问题不知道为什么

log-slave-updates

sync_binlog=1

auto_increment_increment=2

auto_increment_offset=1

进入mysql配置设置为从

 

mysql> change master to master_host='172.16.71.128';

mysql>change master to master_user='repl';
mysql>change master to master_password='ths';

mysql>start slave

主（master）服务器配置：

GRANT REPLICATION SLAVE ON *.* TO 'repl'@'172.16.71.%' IDENTIFIED BY 'ths';

 

 

2）172.16.71.128机器my.cnf配置

skip-name-resolve

server_id           = 1

set-variable = key_buffer_size=512M

set-variable = table_cache=32

set-variable = sort_buffer_size=32M

set-variable = read_buffer_size=32M

set-variable = max_connections=2100

set-variable = log_slow_queries=/var/lib/mysql/log/mysql-slow.log

set-variable = log_bin=/var/lib/mysql/log/mysql-bin

set-variable = long_query_time=1

set-variable = thread_concurrency=16

set-variable = tmp_table_size=4000000000

set-variable = slave_skip_errors=all

#replicate-ignore-db=mysql

#replicate-ignore-db=test

log-slave-updates

sync_binlog=1

auto_increment_increment=2

auto_increment_offset=2

 

进入mysql配置设置为从

 

mysql> change master to master_host='172.16.71.132';

mysql>change master to master_user='repl';
mysql>change master to master_password='ths';

mysql>start slave

主（master）服务器配置：

GRANT REPLICATION SLAVE ON *.* TO 'repl'@'172.16.71.%' IDENTIFIED BY 'ths';

 

五、其它从(slave)服务器配置：

my.cnf配置

skip-name-resolve

server_id           = 1

set-variable = key_buffer_size=512M

set-variable = table_cache=32

set-variable = sort_buffer_size=32M

set-variable = read_buffer_size=32M

set-variable = max_connections=2100

set-variable = log_slow_queries=/var/lib/mysql/log/mysql-slow.log

set-variable = log_bin=/var/lib/mysql/log/mysql-bin

set-variable = long_query_time=1

set-variable = thread_concurrency=16

set-variable = tmp_table_size=4000000000

set-variable = slave_skip_errors=all

 

mysql>change master to master_host='172.16.71.253';
mysql>change master to master_user='repl';
mysql>change master to master_password='ths';                           mysql>change master to master_log_file=’mysql-bin.00001’;
mysql>change master to master_log_pos=128;                                mysql>start slave

或者在my.cnf加入

server-id=2
master-host=172.16.71.253
master-user=repl
master-password=ths

 

问题1：

两台机器互为对方的主，此时不能使用master的VIP作为该两机器的主，如果使用VIP作为master的话没有一台先起来作为vip机器的话些会导致monitor不能使用，日志如下：

2010/12/06 16:29:46  WARN db1 is not replicating from db2

2010/12/06 16:29:46  WARN db2 is not replicating from db1

 

问题2：

当/etc/mysql-mmm/mmm_mon.conf的auto_set_online     3时间设置为较长时，在瞬间关闭现有主，开启现未开启的备时，会导致monitor不能自动切换VIP导致mmm不可用，需要手工设置online状态，有时候甚至会导致monitor启动问题。为防止这情况发生可以把该时间设置短些，或者先让一其中的一台主起来后，再关闭另一台。

 

 

问题3：ERROR：Can't connect to monitor daemon

查看进程一般少了perl相关进程，莫名其妙的感觉，经多次尝试试验，一般都是在mysql那些拥有agent机器上的配置问题，还有monitor的auto_set_online时间设置

[root@localhost ~]# ps axu|grep mmm

root      3113  0.0  7.6  14000  8956 ?        S    16:27   0:00 mmm_mond

root      3115  0.9 33.2 106440 38976 ?        Sl   16:27   0:00 mmm_mond

root      3136  0.4  5.5  10388  6456 ?        S    16:28   0:00 perl /usr/libexec/mysql-mmm/monitor/checker ping_ip

root      3140  0.4  6.5  15208  7640 ?        S    16:28   0:00 perl /usr/libexec/mysql-mmm/monitor/checker mysql

root      3141  0.3  5.4  10388  6448 ?        S    16:28   0:00 perl /usr/libexec/mysql-mmm/monitor/checker ping

root      3142  0.5  6.5  15208  7636 ?        S    16:28   0:00 perl /usr/libexec/mysql-mmm/monitor/checker rep_backlog

root      3144  0.6  6.4  15208  7592 ?        S    16:28   0:00 perl /usr/libexec/mysql-mmm/monitor/checker rep_threads

root      3148  0.0  0.5   3916   652 pts/0    R+   16:28   0:00 grep mmm

 

 

 

六、其它补充资料

主从状态查看：

查看master的状态

SHOW MASTER STATUS;

Position不应为0

查看slave的状态

show slave status;

Slave_IO_Running | Slave_SQL_Running这两个字段 应为YES|YES.

show processlist;查看进程

 

slave start; #启动从复制

STOP SLAVE; #停止从复制
RESET MASTER; #重设master日志
RESET SLAVE;#重设slave日志

 

当Slave_IO_Running: 为No时，从服务器无法同步时操作下如

SLAVE STOP;#停止slave

CHANGE MASTER TO MASTER_LOG_FILE='mysql-bin.000001', MASTER_LOG_POS=0;

#把master_log_file设置为从主show master status;得出来的日志名，POS设置为0

SLAVE START;#再次启用slave

 

 

Mysql普通主从配置（未使用mmm参考）

配置主从my.cnf文件：

skip-innodb

skip-name-resolve

server_id           = 132

set-variable = key_buffer_size=512M

set-variable = table_cache=32

set-variable = sort_buffer_size=32M

set-variable = read_buffer_size=32M

set-variable = max_connections=2100

set-variable = log_slow_queries=/var/lib/mysql/log/mysql-slow.log

set-variable = log_bin=/var/lib/mysql/log/mysql-bin#从库这条语句注释掉

set-variable = long_query_time=1

set-variable = thread_concurrency=16

set-variable = tmp_table_size=4000000000

set-variable = slave_skip_errors=all

因为mysql版本不同在从服务器下开启了下面选项导致同步不能进行

replicate-ignore-db=mysql

replicate-do-db =test

 

 

http://liuyu.blog.51cto.com/183345/98867

http://mysql-mmm.org/mmm2:guide

http://wenku.baidu.com/view/947f0e160b4e767f5acfce37.html

http://blog.chinaunix.net/u3/93755/showart.php?id=2213538