一、InnoDB Cluster介绍
MySQL InnoDB Cluster是MySQL官方推出完整的高可用性解决方案，MySQL InnoDB Cluster由三大组件组成：MySQL Shell，MySQL Router，MySQL Group Replication。每个MySQL服务器实例都运行MySQL Group Replication，它提供了在InnoDB集群内复制数据的机制，具有内置故障转移功能。
MySQL Cluster的部署有两种方式，一种是直接部署MySQL InnoDB Cluster，一种是基于现有的MySQL Group Replication部署，本文采用基于现有的MySQL Group Replication部署
InnoDB Cluster不支持MySQL NDB Cluster
二、环境准备
| 主机名 | IP地址 | 角色 |
|---|---|---|
| oratest51 | 172.16.9.51 | primary |
| oratest52 | 172.16.9.52 | seconde |
| test61 | 172.16.9.61 | seconde |
| node4 | 172.16.8.68 | MySQL Router |
操作系统：CentOS Linux release 7.2.1511
MySQL版本：mysql-5.7.26-linux-glibc2.12-x86_64
MySQL Router版本：mysql-router-8.0.17-linux-glibc2.12-x86_64
MySQL Shell版本：mysql-shell-8.0.17-linux-glibc2.12-x86-64bit
下面的安装步骤前提条件是MGR、MySQL Shell已经安装好的情况下
三、将MGR节点加入MySQL Cluster
3.1登录所有MGR节点，并在该节点的shell中登录所有节点(包括自己)，并保存密码
MySQL JS > shell.connect('root@172.16.9.51:3306')
MySQL JS > shell.connect('root@172.16.9.52:3306')
MySQL JS > shell.connect('root@172.16.9.61:3306')
3.2检查MGR所有节点状态是否正常
MySQL JS > dba.checkInstanceConfiguration('root@172.16.9.52:3306')
Please provide the password for 'root@172.16.9.52:3306': ******
Save password for 'root@172.16.9.52:3306'? [Y]es/[N]o/Ne[v]er (default No): y
Validating MySQL instance at 172.16.9.52:3306 for use in an InnoDB cluster...
This instance reports its own address as oratest52:3306
Clients and other cluster members will communicate with it through this address by default. If this is not correct, the report_host MySQL system variable should be changed.
Checking whether existing tables comply with Group Replication requirements...
No incompatible tables detected
Checking instance configuration...
Instance configuration is compatible with InnoDB cluster
The instance '172.16.9.52:3306' is valid for InnoDB cluster usage.
{
"status": "ok" #状态为ok
}
3.3检查my.cnf配置是否正确，只有正确才能创建集群
MySQL JS > dba.configureLocalInstance('root@172.16.9.51:3306')
The instance '172.16.9.51:3306' belongs to an InnoDB cluster.
Detecting the configuration file...
Default file not found at the standard locations.
Please specify the path to the MySQL configuration file: /etc/my.cnf #输入my.cnf绝对路径
Persisting the cluster settings...
The instance '172.16.9.51:3306' was configured for use in an InnoDB cluster.
The instance cluster settings were successfully persisted.
3.4创建一个名为myCluster的集群
MySQL 172.16.9.51:3306 JS > var cluster = dba.createCluster('myCluster');
A new InnoDB cluster will be created on instance '172.16.9.51:3306'.
Disabling super_read_only mode on instance '172.16.9.51:3306'.
Validating instance at 172.16.9.51:3306...
This instance reports its own address as oratest51:3306
Instance configuration is suitable.
WARNING: Instance '172.16.9.51:3306' cannot persist Group Replication configuration since MySQL version 5.7.26 does not support the SET PERSIST command (MySQL version >= 8.0.11 required). Please use the <Dba>.configureLocalInstance() command locally to persist the chang
es.Creating InnoDB cluster 'myCluster' on '172.16.9.51:3306'...
Adding Seed Instance...
Cluster successfully created. Use Cluster.addInstance() to add MySQL instances.
At least 3 instances are needed for the cluster to be able to withstand up to
one server failure.
3.5查看集群状态，可以发现myCluster集群目前只有一个节点
MySQL 172.16.9.51:3306 JS > cluster.status();
{
"clusterName": "myCluster",
"defaultReplicaSet": {
"name": "default",
"primary": "oratest51:3306",
"ssl": "DISABLED",
"status": "OK_NO_TOLERANCE",
"statusText": "Cluster is NOT tolerant to any failures.",
"topology": {
"oratest51:3306": {
"address": "oratest51:3306",
"mode": "R/W",
"readReplicas": {},
"role": "HA",
"status": "ONLINE"
}
},
"topologyMode": "Single-Primary"
},
"groupInformationSourceMember": "oratest51:3306"
}
3.6添加其他MGR节点到InnoDB Cluster集群当中
以下仅示范一个节点，添加其他节点方法也是一样
MySQL 172.16.9.51:3306 JS > cluster.addInstance('root@oratest52:3306');
Please provide the password for 'root@oratest52:3306': ******
Save password for 'root@oratest52:3306'? [Y]es/[N]o/Ne[v]er (default No): y
The safest and most convenient way to provision a new instance is through
automatic clone provisioning, which will completely overwrite the state of
'oratest52:3306' with a physical snapshot from an existing cluster member. To
use this method by default, set the 'recoveryMethod' option to 'clone'.
The incremental distributed state recovery may be safely used if you are sure
all updates ever executed in the cluster were done with GTIDs enabled, there
are no purged transactions and the new instance contains the same GTID set as
the cluster or a subset of it. To use this method by default, set the
'recoveryMethod' option to 'incremental'.
Incremental distributed state recovery was selected because it seems to be safely usable.
Validating instance at oratest52:3306...
This instance reports its own address as oratest52:3306
Instance configuration is suitable.
WARNING: Instance 'oratest52:3306' cannot persist Group Replication configuration since MySQL version 5.7.26 does not support the SET PERSIST command (MySQL version >= 8.0.11 required). Please use the <Dba>.configureLocalInstance() command locally to persist the changes
.A new instance will be added to the InnoDB cluster. Depending on the amount of
data on the cluster this might take from a few seconds to several hours.
Adding instance to the cluster...
Monitoring recovery process of the new cluster member. Press ^C to stop monitoring and let it continue in background.
State recovery already finished for 'oratest52:3306'
WARNING: Instance 'oratest51:3306' cannot persist configuration since MySQL version 5.7.26 does not support the SET PERSIST command (MySQL version >= 8.0.11 required). Please use the <Dba>.configureLocalInstance() command locally to persist the changes.
The instance 'oratest52:3306' was successfully added to the cluster.
3.7检查InnoDB Cluster集群状态
所有节点都添加到InnoDB Cluster中之后，检查InnoDB Cluster集群状态
MySQL 172.16.9.51:3306 JS > cluster.status();
{
"clusterName": "myCluster",
"defaultReplicaSet": {
"name": "default",
"primary": "oratest51:3306",
"ssl": "DISABLED",
"status": "OK",
"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",
"topology": {
"oratest51:3306": {
"address": "oratest51:3306",
"mode": "R/W",
"readReplicas": {},
"role": "HA",
"status": "ONLINE"
},
"oratest52:3306": {
"address": "oratest52:3306",
"mode": "R/O",
"readReplicas": {},
"role": "HA",
"status": "ONLINE"
},
"test61:3306": {
"address": "test61:3306",
"mode": "R/O",
"readReplicas": {},
"role": "HA",
"status": "ONLINE"
}
},
"topologyMode": "Single-Primary"
},
"groupInformationSourceMember": "oratest51:3306"
}
四、问题汇总
4.1获取不到集群状态，需要先定义集群再获取，如下：
MySQL 172.16.9.51:3306 JS > var cluster = dba.getCluster()
MySQL 172.16.9.51:3306 JS > cluster.status();
4.2节点重启后，如果未加入集群，执行下面代码加入
# mysqlsh --uri root@172.16.9.51:3306
MySQL 172.16.9.51:3306 JS > dba.rebootClusterFromCompleteOutage()
MySQL 172.16.9.51:3306 JS > var cluster = dba.getCluster();
MySQL 172.16.9.51:3306 JS > cluster.rescan()
4.3集群所有节点发生重启
当集群的所有节点都offline，直接获取集群信息失败，如何重新恢复集群
mysql-js> var cluster=dba.getCluster('mycluster')
Dba.getCluster: This function is not available through a session to a standalone instance (RuntimeError)
mysql-js> dba.rebootClusterFromCompleteOutage('mlampCluster'
五、性能测试
5.1在MGR的primary节点插入数据，查看seconde节点的同步情况和性能
5.1.1在MGR的primary节点上插入600W行数据
[root@test51 data]# mysql -uroot -p123456 < db_hq_market_sh_daykline.sql
5.1.2查看seconde节点的各项性能
avg-cpu: %user %nice %system %iowait %steal %idle
4.24 0.00 0.64 1.85 0.00 93.27
Device: tps kB_read/s kB_wrtn/s kB_read kB_wrtn
sda 4.00 0.00 328.80 0 1644
sdb 352.80 0.00 21174.60 0 105873
dm-0 4.80 0.00 328.80 0 1644
dm-1 0.00 0.00 0.00 0 0
avg-cpu: %user %nice %system %iowait %steal %idle
6.62 0.00 0.54 1.78 0.00 91.06
Device: tps kB_read/s kB_wrtn/s kB_read kB_wrtn
sda 0.20 0.00 0.80 0 4
sdb 266.80 0.00 27255.60 0 136278
dm-0 0.20 0.00 0.80 0 4
dm-1 0.00 0.00 0.00 0 0
avg-cpu: %user %nice %system %iowait %steal %idle
2.84 0.00 0.20 1.36 0.00 95.60
Device: tps kB_read/s kB_wrtn/s kB_read kB_wrtn
sda 6.80 0.00 72.00 0 360
sdb 131.60 0.00 18786.10 0 93930
dm-0 7.40 0.00 72.00 0 360
dm-1 0.00 0.00 0.00 0 0
avg-cpu: %user %nice %system %iowait %steal %idle
7.24 0.00 0.73 2.02 0.00 90.02
Device: tps kB_read/s kB_wrtn/s kB_read kB_wrtn
sda 0.40 0.00 148.00 0 740
sdb 316.20 0.00 28787.50 0 143937
dm-0 0.40 0.00 148.00 0 740
dm-1 0.00 0.00 0.00 0 0
avg-cpu: %user %nice %system %iowait %steal %idle
3.44 0.00 0.36 1.01 0.00 95.18
Device: tps kB_read/s kB_wrtn/s kB_read kB_wrtn
sda 4.20 12.80 39.00 64 195
sdb 145.40 0.00 15742.60 0 68713
dm-0 5.40 12.80 39.00 64 195
dm-1 0.00 0.00 0.00 0 0
mysql> select count(*) from t_daykline;
+----------+
| count(*) |
+----------+
| 6088654 |
+----------+
测试结果： IOPS基本维持在100到350之间，每秒写入基本在15M~22M之间，主库数据导入成功后，从库数据也实时同步完成
六、个人总结
- MGR架构可以实现高可用，但实现failover，则需要安装InnoDB Cluster了。MySQL Shell和MySQL Router是InnoDB Cluster集群的一部分。MySQL Shell用于安装和管理InnoDB Cluster集群，InnoDB Cluster集群通过新建的原数据库
MySQL_innodb_cluster_metadata管理MGR集群状态，MySQL Router通过周期性的访问InnoDB Cluster创建的MySQL_innodb_cluster_metadata库中的元数据获取集群成员信息,再通过performance_schema的系统表获取可连接实例及其状态来实现failover和读负载均衡、读写分离的路由。 - 创建InnoDB Cluster至少需要三台MySQL服务器，且数量为奇数个。
- MySQL InnoDB Cluster作为MySQL官方推荐的两个集群解决方案之一，个人觉得各方面还是不错的，对功能要求不多的话可以选择此方案，第三方的高可用方案如：HAProxy、Proxsql也可以。