CentOS7.3下Percona-XtraDB-Cluster-5.7.19集群部署
PXC三节点安装：
node1：10.10.10.11
node2：10.10.10.12
node3：10.10.10.13
软件版本选择：
Percona-XtraDB-Cluster-5.7.19-rel17-29.22.3.Linux.x86_64.ssl101
percona-xtrabackup-2.4.11-Linux-x86_64.libgcrypt11.tar.gz/percona-xtrabackup-2.4.12-Linux-x86_64.libgcrypt11.tar.gz
注意openssl版本,ssl100、ssl101、ssl102分别对应三类linux版本，所以下载ssl101
#node1、node2、node3
操作系统版本
[root@node1 ~]# cat /etc/redhat-release
CentOS Linux release 7.3.1611 (Core)
[root@node1 ~]# systemctl status firewalld.service
* firewalld.service - firewalld - dynamic firewall daemon
Loaded: loaded (/usr/lib/systemd/system/firewalld.service; disabled; vendor preset: enabled)
Active: inactive (dead)
Docs: man:firewalld(1)
[root@node1 ~]#
[root@node1 ~]# getenforce
Disabled
[root@node1 ~]#cat /etc/hosts
#for pxc test
10.10.10.13 node3
10.10.10.11 node1
10.10.10.12 node2
安装依赖包：
yum install -y git scons gcc gcc-c++ openssl check cmake bison boost-devel asio-devel libaio-devel ncurses-devel readline-devel pam-devel socat
创建MySQL,basedir和datadir,编辑my.cnf
[root@node1 ~]#ls /opt/source
percona-xtrabackup-2.4.12-Linux-x86_64.libgcrypt11.tar.gz
Percona-XtraDB-Cluster-5.7.19-rel17-29.22.3.Linux.x86_64.ssl101.tar.gz
[root@node1 ~]#tar -zxf percona-xtrabackup-2.4.12-Linux-x86_64.libgcrypt11.tar.gz
[root@node1 ~]#tar -zxf Percona-XtraDB-Cluster-5.7.19-rel17-29.22.3.Linux.x86_64.ssl101.tar.gz
[root@node1 ~]#ln -s /opt/source/Percona-XtraDB-Cluster-5.7.14-rel8-26.17.1.Linux.x86_64.ssl100 /usr/local/pxc
ln -s /opt/source/xtrabackup-2.4.11/bin/xtrabackup /opt/source/Percona-XtraDB-Cluster-5.7.22-rel22-29.26.1.Linux.x86_64.ssl101/bin/xtrabackup
[root@node1 ~]#mkdir -p /data/mysql_33071/{data,logs,run,tmp}
[root@node1 ~]#groupadd -g 1000 mysql
[root@node1 ~]#useradd -r -g mysql -u 1000 -s /bin/false mysql
编辑my.cnf配置文件
[root@node1 ~]#vi /data/mysql_33071/my.cnf
[root@node1 ~]#chown -R mysql:mysql /data/mysql_33071
[root@node1 ~]#chown -R mysql:mysql /usr/local/pxc
node1的操作：
#node1:初始化实例和启动实例方式
/usr/local/pxc/bin/mysqld --defaults-file=/data/mysql_33071/my.cnf --initialize --basedir=/usr/local/pxc --datadir=/data/mysql_33071/data --user=mysql
在错误日志，记录下root初始密码！
启动实例：
/usr/local/pxc/bin/mysqld_safe --defaults-file=/data/mysql_33071/my.cnf --ledir=/usr/local/pxc/bin --wsrep_new_cluster &
#当采用自定义路径的my.cnf初始化及启动实例的时候，需要首先指定defaults-file；而且mysqld采用自定义路径时，需要指定ledir
[root@masterdb ~]#ps -ef |grep mysql
root 27690 1 0 9 ? 00:00:00 /bin/sh /usr/local/pxc/bin/mysqld_safe --defaults-file=/data/mysql_33071/my.cnf --ledir=/usr/local/pxc/bin --wsrep_new_cluster
mysql 28111 27690 0 9 ? 00:11:07 /usr/local/pxc/bin/mysqld --defaults-file=/data/mysql_33071/my.cnf --basedir=/usr/local/pxc --datadir=/data/mysql_33071/data
--plugin-dir=/usr/local/pxc/lib/mysql/plugin --user=mysql --wsrep-provider=/usr/local/pxc/lib/libgalera_smm.so --wsrep-new-cluster --log-error=/data/mysql_33071/logs/mysql_error.log --pid-file=/data/mysql_33071/logs/mysql.pid
--socket=/data/mysql_33071/run/mysql.sock --port=33071 --wsrep_start_position=4946b0f0-ad00-11e8-8524-f2562c72b43b:5
修改root密码：
[root@node1 ~]#mysql -uroot -p -S /data/mysql_33071/run/mysql.sock
mysql> set password=password('root');
mysql> flush privileges;
创建sst user，并授予相应权限
mysql> CREATE USER 'sstuser'@'localhost' IDENTIFIED BY 'sstuser';
mysql> GRANT RELOAD, LOCK TABLES, PROCESS, REPLICATION CLIENT ON *.* TO 'sstuser'@'localhost';
mysql> FLUSH PRIVILEGES;
备份数据库
[root@node1 ~]#innobackupex --user=root --password="root" --socket=/data/mysql_33071/run/mysql.sock /tmp/
node2、node3的操作：
###node2、node3：###
用node1的备份，在节点2、节点3进行恢复;启动，加入复制集
innobackupex --defaults-file=/etc/my.cnf --user=root --password=xxxxxx --apply-log /data/2019-01-12_22-45-44
innobackupex --defaults-file=/etc/my.cnf --user=root --password=xxxxxx --copy-back /data/2019-01-12_22-45-44
/usr/local/pxc/bin/mysqld_safe --defaults-file=/data/mysql_33071/my.cnf --ledir=/usr/local/pxc/bin &
验证：
node1检查pxc状态
mysql> show global status like 'wsrep_cluster_size';
+--------------------+-------+
| Variable_name | Value |
+--------------------+-------+
| wsrep_cluster_size | 3 |
+--------------------+-------+
1 row in set (0.01 sec)
show status like 'wsrep%';
mysql> show global status like 'wsrep_incoming_addresses';
+--------------------------+-----------------------------------------------------------+
| Variable_name | Value |
+--------------------------+-----------------------------------------------------------+
| wsrep_incoming_addresses | 10.10.10.12:33071,10.10.10.13:33071,10.10.10.11:33071 |
+--------------------------+-----------------------------------------------------------+
1 row in set (0.01 sec)
[root@node1 ~]#vi /data/mysql_33071/my.cnf
[client]
socket=/data/mysql_33071/run/mysql.sock
port = [mysqld]
port =
server_id =
user = mysql
basedir = /usr/local/pxc
datadir =/data/mysql_33071/data
tmpdir = /data/mysql_33071/tmp socket = /data/mysql_33071/run/mysql.sock
pid-file = /data/mysql_33071/logs/mysql.pid
log-error = /data/mysql_33071/logs/mysql_error.log
slow_query_log_file = /data/mysql_33071/logs/slow.log
log_warnings =
long_query_time =
sql_mode=NO_ENGINE_SUBSTITUTION,STRICT_TRANS_TABLES innodb_buffer_pool_size = 2048M ## pxc variables ##
# Path to Galera library
wsrep_provider=/usr/local/pxc/lib/libgalera_smm.so
# Cluster connection URL contains the IPs of node#, node# and node#
wsrep_cluster_address=gcomm://10.10.10.11,10.10.10.12,10.10.10.13
# In order for Galera to work correctly binlog format should be ROW
binlog_format=ROW
# MyISAM storage engine has only experimental support
default_storage_engine=InnoDB
# This InnoDB autoincrement locking mode is a requirement for Galera
innodb_autoinc_lock_mode=
# Node name
#wsrep_node_name = pxc1
wsrep_node_name = masterdb
# Node address
wsrep_node_address=10.10.10.11
# SST method默认情况下，Percona XtraDB群集使用Percona XtraBackup进行状态快照传输（SST）
wsrep_sst_method=xtrabackup-v2
# Cluster name
wsrep_cluster_name=my-pxc
# Authentication for SST method
wsrep_sst_auth="sstuser:sstuser" #将SST的身份验证凭据指定为<sst_user>：<sst_pass>
node1:my.cnf
[client]
socket=/data/mysql_33071/run/mysql.sock
port = [mysqld]
port =
server_id =
user = mysql
basedir = /usr/local/pxc
datadir =/data/mysql_33071/data
tmpdir = /data/mysql_33071/tmp socket = /data/mysql_33071/run/mysql.sock
pid-file = /data/mysql_33071/logs/mysql.pid
log-error = /data/mysql_33071/logs/mysql_error.log
slow_query_log_file = /data/mysql_33071/logs/slow.log
log_warnings =
long_query_time =
sql_mode=NO_ENGINE_SUBSTITUTION,STRICT_TRANS_TABLES innodb_buffer_pool_size = 2048M ## pxc variables ##
# Path to Galera library
wsrep_provider=/usr/local/pxc/lib/libgalera_smm.so
# Cluster connection URL contains the IPs of node#, node# and node#
wsrep_cluster_address=gcomm://10.10.10.11,10.10.10.12,10.10.10.13
# In order for Galera to work correctly binlog format should be ROW
binlog_format=ROW
# MyISAM storage engine has only experimental support
default_storage_engine=InnoDB
# This InnoDB autoincrement locking mode is a requirement for Galera
innodb_autoinc_lock_mode=
# Node name
#wsrep_node_name = pxc2
wsrep_node_name = qa06..yz
# Node address
wsrep_node_address=10.10.10.12
# SST method
wsrep_sst_method=xtrabackup-v2
# Cluster name
wsrep_cluster_name=my-pxc
# Authentication for SST method
wsrep_sst_auth="sstuser:sstuser" pxc_strict_mode=ENFORCING
node2:my.cnf
[client]
socket=/data/mysql_33071/run/mysql.sock
port = [mysqld]
port =
server_id =
user = mysql
basedir = /usr/local/pxc
datadir =/data/mysql_33071/data
tmpdir = /data/mysql_33071/tmp socket = /data/mysql_33071/run/mysql.sock
pid-file = /data/mysql_33071/logs/mysql.pid
log-error = /data/mysql_33071/logs/mysql_error.log
slow_query_log_file = /data/mysql_33071/logs/slow.log
log_warnings =
long_query_time =
sql_mode=NO_ENGINE_SUBSTITUTION,STRICT_TRANS_TABLES innodb_buffer_pool_size = 2048M ## pxc variables ##
# Path to Galera library
wsrep_provider=/usr/local/pxc/lib/libgalera_smm.so
# Cluster connection URL contains the IPs of node#, node# and node#
wsrep_cluster_address=gcomm://10.10.10.11,10.10.10.12,10.10.10.13
# In order for Galera to work correctly binlog format should be ROW
binlog_format=ROW
# MyISAM storage engine has only experimental support
default_storage_engine=InnoDB
# This InnoDB autoincrement locking mode is a requirement for Galera
innodb_autoinc_lock_mode=
# Node name
wsrep_node_name = cgdb..yz
# Node address
wsrep_node_address=10.10.10.13
# SST method
wsrep_sst_method=xtrabackup-v2
# Cluster name
wsrep_cluster_name=my-pxc
# Authentication for SST method
wsrep_sst_auth="sstuser:sstuser" pxc_strict_mode=ENFORCING
node3:my.cnf
Percona XtraDB Cluster与MySQL Replication区别在于：
分布式系统的CAP理论：
C—一致性，所有节点的数据一致；
A—可用性，一个或多个节点失效，不影响服务请求；
P—分区容忍性，节点间的连接失效，仍然可以处理请求；
任何一个分布式系统，需要满足这三个中的两个。
MySQL Replication: 可用性和分区容忍性；
Percona XtraDBCluster: 一致性和可用性。
因此MySQL Replication并不保证数据的一致性，而Percona XtraDB Cluster提供数据一致性
阅读：
https://www.cnblogs.com/kevingrace/p/5685371.html