pacemaker+corosync实现mysql的高可用
实现原理
创建集群，添加节点，添加共享磁盘进行资源共享，将共享磁盘挂载在mysql的目录上/var/lib/mysql/，就是哪一个节点使用资源进行挂载，哪一个节点可以使用该数据库。
[root@toto1 ~]# ls /var/lib/mysql/
[root@toto1 ~]# systemctl start mariadb.service
[root@toto1 ~]# ls /var/lib/mysql/
aria_log.00000001 ibdata1 ib_logfile1 mysql.sock test
aria_log_control ib_logfile0 mysql performance_schema
[root@toto1 ~]# systemctl stop mariadb.service
[root@toto1 ~]# ls /var/lib/mysql/
aria_log.00000001 ibdata1 ib_logfile1 performance_schema
aria_log_control ib_logfile0 mysql test
1 、作为集群节点的服务器全部安装mariadb
yum install mariadb-server.x86_64 -y
2 、增加资源
vip：#增加VIP资源
pcs resource create totovip1 ocf:heartbeat:IPaddr2 ip=172.25.13.100 cidr_netmask=32 op monitor interval=30s
mysql_data：#增加文件管理系统资源
pcs resource create mysql_data ocf:heartbeat:Filesystem device=/dev/sdb1 directory=/var/lib/mysql fstype=xfs op monitor interval=30s
mariadb：#增加数据库服务
pcs resource create mariadb systemd:mariadb op monitor interval=1min
mysql_group：将资源添加到mysql_group组中
pcs resource group add mysql_group mysql_data totovip1 mariadb
测试：
数据库资源运行在节点toto2上
[root@toto1 ~]# pcs status
Cluster name: toto_cluster
Stack: corosync
Current DC: toto1 (version 1.1.15-11.el7-e174ec8) - partition with quorum
Last updated: Wed Jun 26 00:14:39 2019 Last change: Wed Jun 26 00:09:25 2019 by root via cibadmin on toto1
2 nodes and 4 resources configured
Online: [ toto1 toto2 ]
Full list of resources:
vmfence (stonith:fence_xvm): Started toto1
Resource Group: mysql_group
mysql_data (ocf::heartbeat:Filesystem): Started toto2
totovip1 (ocf::heartbeat:IPaddr2): Started toto2
mariadb (systemd:mariadb): Started toto2 # 资源运行在节点toto2上
查看各个节点共享磁盘挂载情况：
toto1节点：共享磁盘没有挂载，数据库没有启动
[root@toto1 ~]# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/rhel-root 17811456 1272992 16538464 8% /
devtmpfs 497292 0 497292 0% /dev
tmpfs 508264 55056 453208 11% /dev/shm
tmpfs 508264 13088 495176 3% /run
tmpfs 508264 0 508264 0% /sys/fs/cgroup
/dev/sda1 1038336 141512 896824 14% /boot
tmpfs 101656 0 101656 0% /run/user/0
toto2 节点：共享磁盘挂载在/var/lib/mysql/ 数据库启动
[root@toto2 ~]# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/sdb1 20960256 62952 20897304 1% /var/lib/mysql
当节点toto2异常导致资源转移到节点toto1 的时候。共享磁盘在节点toto1上挂载，数据库在节点toto1上运行。
[root@toto1 ~]# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/sdb1 20960256 62952 20897304 1% /var/lib/mysql # 共享磁盘挂载
[root@toto1 ~]# systemctl status mariadb.service # 服务自动开启
Active: active (running) since Wed 2019-06-26 00:19:24 CST; 19s ago
[root@toto1 ~]# mysql # 数据库可以正常使用
Welcome to the MariaDB monitor. Commands end with ; or \g.
MariaDB [(none)]> show databases;
+--------------------+
| Database |
+--------------------+
| information_schema |
| mysql |
| performance_schema |
| test |
+--------------------+
4 rows in set (0.00 sec)
MariaDB [(none)]> quit
Bye
到次，mysql服务的高可用已经实现
现象：资源目前在节点toto2，当toto2宕机之后，资源会暂时回到toto1上，但是当toto2重启以后，资源就又回到了toto2上
但是可以看出这个资源是不正确的，资源应该在重启之后在不进行转移，因为我们并没有定义错误域，服务资源应该在toto1上，fence应该在toto2上。但是由于这个资源是具有粘滞性，导致出现上述异常。此时可以对其进行解决。方法步骤如下:
1 、删除之前创建资源、组、fence
pcs resource delete mysql_group
pcs resource delete vmfence
pcs property set stonith-enabled=false
2 、重现提阿加资源、组
pcs resource create totovip1 ocf:heartbeat:IPaddr2 ip=172.25.13.100 cidr_netmask=32 op monitor interval=30s
pcs resource create mysql_data ocf:heartbeat:Filesystem device=/dev/sdb1 directory=/var/lib/mysql fstype=xfs op monitor interval=30s
pcs resource create mariadb systemd:mariadb op monitor interval=1min
pcs resource group add mysql_group mysql_data totovip1 mariadb
3 、更改resource-stickiness值=100
[root@toto1 ~]# pcs resource defaults resource-stickiness=100
[root@toto1 ~]# pcs resource defaults
resource-stickiness: 100
4 、重新添加fence并再次更改resource-stickiness值=0
[root@toto1 ~]# pcs stonith create vmfence fence_xvm pcmk_host_map="toto1:toto1;toto2:toto2" op monitor interval=1min
[root@toto1 ~]# pcs resource defaults resource-stickiness=0
[root@toto1 ~]# pcs resource defaults
resource-stickiness: 0
5 、设置启用stonish
pcs property set stonith-enabled=true
测试
[root@toto2 ~]# echo c >/proc/sysrq-trigger
可以看出粘滞性问题已经解决。重新开启服务之后fence改变，资源不变