mysql主主、mysql-proxy讀寫分離、keepalived高可用完美組合

配置mysql主主複製

服務器A：172.16.100.9
服務器B：172.16.100.10

首先在服務器A和服務器B上各安裝好單機版mysql5.7，安裝好後，再繼續下面的操作。

在服務器A和服務器B上都執行

wget ftp://ftp.pbone.net/mirror/ftp.centos.org/7.6.1810/cloud/x86_64/openstack-ocata/common/crudini-0.9-2.el7.noarch.rpm
wget ftp://ftp.pbone.net/mirror/ftp.centos.org/7.6.1810/cloud/x86_64/openstack-stein/openstack-utils-2017.1-1.el7.noarch.rpm
rpm -ivh crudini-0.9-2.el7.noarch.rpm
rpm -ivh openstack-utils-2017.1-1.el7.noarch.rpm

mkdir -p /usr/local/mysqldata/binlogs
mkdir -p /usr/local/mysqldata/relaylogs
chown -R mysql:mysql  /usr/local/mysqldata
chmod -R 770  /usr/local/mysqldata

1.控制自增列的初始值爲2和增量步長爲1

openstack-config --set /etc/my.cnf mysqld auto_increment_offset 1
openstack-config --set /etc/my.cnf mysqld auto_increment_increment 2

2.啓用binlog日誌，只保留7天內的日誌

openstack-config --set /etc/my.cnf mysqld log-bin /usr/local/mysqldata/binlogs/mysql-bin
openstack-config --set /etc/my.cnf mysqld max_binlog_size  1G
openstack-config --set /etc/my.cnf mysqld binlog_format ROW
openstack-config --set /etc/my.cnf mysqld expire_logs_days 7

3.配置中繼日誌的位置

openstack-config --set /etc/my.cnf mysqld relay-log /usr/local/mysqldata/relaylogs/slave-relay-bin

4.開啓從庫上的更新操作要寫入binlog日誌，以便從庫的從庫能夠進行日誌同步

openstack-config --set /etc/my.cnf mysqld log-slave-updates

5.設置每緩存20個事務就採用同步寫磁盤的方式寫入binlog日誌

openstack-config --set /etc/my.cnf mysqld innodb_support_xa 1
openstack-config --set /etc/my.cnf mysqld sync_binlog 20

6.不記錄系統數據庫的二進制日誌

sed -i "/^\[mysqld/a\binlog_ignore_db = mysql" /etc/my.cnf
sed -i "/^\[mysqld/a\binlog_ignore_db = information_schema" /etc/my.cnf
sed -i "/^\[mysqld/a\binlog_ignore_db = information_schema" /etc/my.cnf
sed -i "/^\[mysqld/a\binlog_ignore_db = sys" /etc/my.cnf

7.設置服務器的server-id

openstack-config --set /etc/my.cnf mysqld server-id $RANDOM

8.重啓數據庫

ps -ef | grep mysql | grep -v grep | awk '{print $2}' |xargs kill -9
/usr/local/mysqlrun/bin/mysqld_safe --defaults-file=/etc/my.cnf &

9.創建同步賬號

grant replication slave on *.* to 'repl'@'172.16.100.9' identified by 'repl1234!';
grant replication slave on *.* to 'repl'@'172.16.100.10' identified by 'repl1234!';
flush privileges;

10.查看並記錄master狀態中的File和Position

reset master;
show master status;
得到的結果爲：
mysql-bin.000001 和 154

在服務器A上執行

stop slave;
reset slave all;
change master to master_host='172.16.100.10',master_user='repl',master_password='repl1234!',master_log_file='mysql-bin.000001',master_log_pos=154;
start slave;
show slave status\G;

在服務器B上執行

stop slave;
reset slave all;
change master to master_host='172.16.100.9',master_user='repl',master_password='repl1234!',master_log_file='mysql-bin.000001',master_log_pos=154;
start slave;
show slave status\G;

 

配置keepalived高可用(不搶佔)

服務器A：172.16.100.9
服務器B：172.16.100.10
VIP：172.16.100.12
VIP：172.16.100.13

首先在服務器A和服務器B上各安裝好keepalived，安裝好後，再繼續下面的操作。

在服務器A上執行

1.編寫腳本監控mysql狀態

echo '#!/bin/bash
count = `ps aux | grep -v grep | grep mysql | wc -l`
if [ $count > 0 ];then
exit 0
else
exit 1
fi' > /etc/keepalived/checkmysql.sh

2.配置全局定義塊

global_defs {
     router_id mysqlserver01
}

3.keepalived只能做到對網絡故障和keepalived本身的監控,配置腳本監控mysql服務狀態

weight對priority的影響詳情請參考https://www.cnblogs.com/arjenlee/p/9258188.html
vrrp_script checkmysql
{
     script "/etc/keepalived/checkmysql.sh"
     interval 3
     weight -20
}

4.配置VRRP實例定義塊

vrrp_instance instance1 {
     state MASTER
     virtual_router_id 1
     interface eth0
     mcast_src_ip 172.16.100.9
     priority 100
     advert_int 3
     nopreempt
     authentication {
         auth_type PASS
         auth_pass kee1234
     }
     virtual_ipaddress {
     172.16.100.12
     }
     track_script {
     checkmysql
     }
}
vrrp_instance instance2 {
     state BACKUP
     virtual_router_id 2
     interface eth0
     mcast_src_ip 172.16.100.9
     priority 90
     advert_int 3
     nopreempt
     authentication {
         auth_type PASS
         auth_pass kee1234
     }
     virtual_ipaddress {
     172.16.100.13
     }
     track_script {
     checkmysql
     }
}

在服務器B上執行

1.編寫腳本監控mysql狀態

echo '#!/bin/bash
count = `ps aux | grep -v grep | grep mysql | wc -l`
if [ $count > 0 ];then
exit 0
else
exit 1
fi' > /etc/keepalived/checkmysql.sh

2.配置全局定義塊

global_defs {
     router_id mysqlserver02
}

3.keepalived只能做到對網絡故障和keepalived本身的監控,配置腳本監控mysql服務狀態

weight對priority的影響詳情請參考https://www.cnblogs.com/arjenlee/p/9258188.html
vrrp_script checkmysql
{
     script "/etc/keepalived/checkmysql.sh"
     interval 3
     weight -20
}

4.配置VRRP實例定義塊

vrrp_instance instance1 {
     state BACKUP
     virtual_router_id 1
     interface eth0
     mcast_src_ip 172.16.100.10
     priority 90
     advert_int 3
     nopreempt
     authentication {
         auth_type PASS
         auth_pass kee1234
     }
     virtual_ipaddress {
     172.16.100.12
     }
     track_script {
     checkmysql
     }
}
vrrp_instance instance2 {
     state MASTER
     virtual_router_id 2
     interface eth0
     mcast_src_ip 172.16.100.10
     priority 100
     advert_int 3
     nopreempt
     authentication {
         auth_type PASS
         auth_pass kee1234
     }
     virtual_ipaddress {
     172.16.100.13
     }
     track_script {
     checkmysql
     }
}

mysql-proxy實現讀寫分離

服務器：172.16.100.7

首先在服務器上安裝好mysql-proxy，安裝好後，再繼續下面的操作。

1.修改配置文件

cat << EOF > /etc/mysql-proxy.cnf
[mysql-proxy]
user = root
admin-username=root
admin-password=123456
proxy-address=172.16.100.7
proxy-lua-script = /usr/local/myapps/mysql-proxy/lua/rw-splitting.lua
admin-lua-script = /usr/local/myapps/mysql-proxy/lua/admin-sql.lua
proxy-backend-addresses = 172.16.100.12
proxy-read-only-backend-addresses = 172.16.100.13
log-file=/usr/local/myapps/mysql-proxy/logs/mysql-proxy.log
log-level=debug
EOF

2.重啓mysql-proxy服務

ps -ef | grep mysql-proxy | grep -v grep | awk '{print $2}' |xargs kill -9
/usr/local/myapps/mysql-proxy/bin/mysql-proxy --daemon --keepalive  --defaults-file=/etc/mysql-proxy.cnf