MHA+ProxySQL实现读写分离高可用 - yayun - 博客园

标签： | 发表时间：2018-09-28 10:43 | 作者：
出处：https://www.cnblogs.com

最近在研究ProxySQL，觉得还挺不错的，所以就简单的折腾了一下，ProxySQL目前也是Percona在推荐的一个读写分离的中间件。关于详细的介绍可以参考官方文档。 https://github.com/sysown/proxysql/wiki

本文主要介绍的是MHA+ProxySQL读写分离以及高可用，ProxySQL的细节请参考文档，目前已经有人写的非常详细了，文章最后会给出链接。当然和Group Replication，PXC搭配那是更完美的。关于MHA的配置参考我前面的文章，本文就不再介绍。下面来看看

MHA和ProxySQL搭配使用的架构图：

 

测试环境（1主2从）：

M-> 192.168.0.20
S1-> 192.168.0.30
S2-> 192.168.0.40
proxysql -> 192.168.0.10

1. 检查主从复制是否正常：

[root@db_server_yayun_04 ~]# masterha_check_repl --conf=/data/mha/3306/mha.cnf 
Thu Jun 15 17:45:32 2017 - [warning] Global configuration file /etc/masterha_default.cnf not found. Skipping.
Thu Jun 15 17:45:32 2017 - [info] Reading application default configuration from /data/mha/3306/mha.cnf..
Thu Jun 15 17:45:32 2017 - [info] Updating application default configuration from /usr/local/bin/load_cnf..
Thu Jun 15 17:45:32 2017 - [info] Reading server configuration from /data/mha/3306/mha.cnf..
Thu Jun 15 17:45:32 2017 - [info] Setting max_ping_errors to 10, ping_interval to 3.
Thu Jun 15 17:45:32 2017 - [info] MHA::MasterMonitor version 0.57.
Thu Jun 15 17:45:32 2017 - [info] GTID failover mode = 1
Thu Jun 15 17:45:32 2017 - [info] Dead Servers:
Thu Jun 15 17:45:32 2017 - [info] Alive Servers:
Thu Jun 15 17:45:32 2017 - [info]   192.168.0.20(192.168.0.20:3306)
Thu Jun 15 17:45:32 2017 - [info]   192.168.0.30(192.168.0.30:3306)
Thu Jun 15 17:45:32 2017 - [info]   192.168.0.40(192.168.0.40:3306)
Thu Jun 15 17:45:32 2017 - [info] Alive Slaves:
Thu Jun 15 17:45:32 2017 - [info]   192.168.0.30(192.168.0.30:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Thu Jun 15 17:45:32 2017 - [info]     GTID ON
Thu Jun 15 17:45:32 2017 - [info]     Replicating from 192.168.0.20(192.168.0.20:3306)
Thu Jun 15 17:45:32 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Thu Jun 15 17:45:32 2017 - [info]   192.168.0.40(192.168.0.40:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Thu Jun 15 17:45:32 2017 - [info]     GTID ON
Thu Jun 15 17:45:32 2017 - [info]     Replicating from 192.168.0.20(192.168.0.20:3306)
Thu Jun 15 17:45:32 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Thu Jun 15 17:45:32 2017 - [info] Current Alive Master: 192.168.0.20(192.168.0.20:3306)
Thu Jun 15 17:45:32 2017 - [info] Checking slave configurations..
Thu Jun 15 17:45:32 2017 - [info] Checking replication filtering settings..
Thu Jun 15 17:45:32 2017 - [info]  binlog_do_db= , binlog_ignore_db= 
Thu Jun 15 17:45:32 2017 - [info]  Replication filtering check ok.
Thu Jun 15 17:45:32 2017 - [info] GTID (with auto-pos) is supported. Skipping all SSH and Node package checking.
Thu Jun 15 17:45:32 2017 - [info] Checking SSH publickey authentication settings on the current master..
Thu Jun 15 17:45:33 2017 - [info] HealthCheck: SSH to 192.168.0.20 is reachable.
Thu Jun 15 17:45:33 2017 - [info] 
192.168.0.20(192.168.0.20:3306) (current master)
 +--192.168.0.30(192.168.0.30:3306)
 +--192.168.0.40(192.168.0.40:3306)

Thu Jun 15 17:45:33 2017 - [info] Checking replication health on 192.168.0.30..
Thu Jun 15 17:45:33 2017 - [info]  ok.
Thu Jun 15 17:45:33 2017 - [info] Checking replication health on 192.168.0.40..
Thu Jun 15 17:45:33 2017 - [info]  ok.
Thu Jun 15 17:45:33 2017 - [warning] master_ip_failover_script is not defined.
Thu Jun 15 17:45:33 2017 - [warning] shutdown_script is not defined.
Thu Jun 15 17:45:33 2017 - [info] Got exit code 0 (Not master dead).

MySQL Replication Health is OK.
[root@db_server_yayun_04 ~]#

2. 配置后端MySQL。登入ProxySQL，把MySQL主从的信息添加进去。将主库master也就是做写入的节点放到HG 100中，salve节点做读放到HG 1000。在proxysql输入命令：

mysql -uadmin -padmin -h127.0.0.1 -P6032
[[email protected]][(none)]> insert into mysql_servers(hostgroup_id,hostname,port,weight,max_connections,max_replication_lag,comment) values(100,'192.168.0.20',3306,1,1000,10,'test my proxysql');
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]> insert into mysql_servers(hostgroup_id,hostname,port,weight,max_connections,max_replication_lag,comment) values(1000,'192.168.0.30',3306,1,1000,10,'test my proxysql');
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]> insert into mysql_servers(hostgroup_id,hostname,port,weight,max_connections,max_replication_lag,comment) values(1000,'192.168.0.40',3306,1,1000,10,'test my proxysql'); 
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]> select * from mysql_servers;
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
| hostgroup_id | hostname     | port | status | weight | compression | max_connections | max_replication_lag | use_ssl | max_latency_ms | comment          |
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
| 100          | 192.168.0.20 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 0              | test my proxysql |
| 1000         | 192.168.0.30 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 0              | test my proxysql |
| 1000         | 192.168.0.40 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 0              | test my proxysql |
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
3 rows in set (0.00 sec)

3. 配置后端MySQL用户。这个用户需要先在后端MySQL(20,30,40)里真实存在，一个是监控账号、一个是程序账号。
(1) 监控账号（用来监控后端mysql是否存活以及read_only变量）：

GRANT SUPER, REPLICATION CLIENT ON *.* TO 'proxysql'@'192.168.0.10' IDENTIFIED BY 'proxysql';

(2) 程序账号（这里为了后面测试方便给了all权限）：

GRANT all ON *.* TO 'yayun'@'192.168.0.10' identified by 'yayun';

4. 在后端MySQL里添加完之后再配置ProxySQL：这里需要注意，default_hostgroup需要和上面的对应。（proxysql）

[[email protected]][(none)]> insert into mysql_users(username,password,active,default_hostgroup,transaction_persistent) values('yayun','yayun',1,100,1);
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]> select * from mysql_users;
+----------+----------+--------+---------+-------------------+----------------+---------------+------------------------+--------------+---------+----------+-----------------+
| username | password | active | use_ssl | default_hostgroup | default_schema | schema_locked | transaction_persistent | fast_forward | backend | frontend | max_connections |
+----------+----------+--------+---------+-------------------+----------------+---------------+------------------------+--------------+---------+----------+-----------------+
| yayun    | yayun    | 1      | 0       | 100               | NULL           | 0             | 1                      | 0            | 1       | 1        | 10000           |
+----------+----------+--------+---------+-------------------+----------------+---------------+------------------------+--------------+---------+----------+-----------------+
1 row in set (0.00 sec)

[[email protected]][(none)]>

5. 设置健康监测账号（proxysql）：

[[email protected]][(none)]> set mysql-monitor_username='proxysql';
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]> set mysql-monitor_password='proxysql';
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]>

6. 加载配置和变量：因为修改了servers、users和variables，所以加载的时候要执行：

[[email protected]][(none)]> load mysql servers to runtime;
Query OK, 0 rows affected (0.01 sec)

[[email protected]][(none)]> load mysql users to runtime;
Query OK, 0 rows affected (0.00 sec)

[[email protected]][(none)]> load mysql variables to runtime;
Query OK, 0 rows affected (0.00 sec)

[[email protected]][(none)]> save mysql servers to disk;
Query OK, 0 rows affected (0.05 sec)

[[email protected]][(none)]> save mysql users to disk;
Query OK, 0 rows affected (0.02 sec)

[[email protected]][(none)]> save mysql variables to disk;
Query OK, 74 rows affected (0.01 sec)

[[email protected]][(none)]>

7. 连接数据库，通过proxysql的客户端接口访问(6033)：(我这里从40从库上面发起连接)

 mysql -uyayun -pyayun -h192.168.0.10 -P6033
[root@db_server_yayun_04 ~]# mysql -uyayun -pyayun -h192.168.0.10 -P6033               
mysql: [Warning] Using a password on the command line interface can be insecure.
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 4
Server version: 5.5.30 (ProxySQL)

Copyright (c) 2000, 2016, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.

Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

[[email protected]][(none)]> show databases;
+--------------------+
| Database           |
+--------------------+
| information_schema |
| mysql              |
| performance_schema |
| sys                |
| yayun              |
+--------------------+
5 rows in set (0.01 sec)

[[email protected]][(none)]>

 8. 创建表并且写入数据进行查询：

[[email protected]][(none)]> use yayun
Database changed, 2 warnings
[[email protected]][yayun]> create table t1 ( id int);
Query OK, 0 rows affected (0.08 sec)

[[email protected]][yayun]> insert into t1 select 1;
Query OK, 1 row affected (0.05 sec)
Records: 1  Duplicates: 0  Warnings: 0

[[email protected]][yayun]> select * from t1;
+------+
| id   |
+------+
|    1 |
+------+
1 row in set (0.00 sec)

[[email protected]][yayun]>

可以看到创建了表，并且插入了数据，查询也没问题。proxysql有个类似审计的功能，可以查看各类SQL的执行情况。在proxysql执行SQL查看。

[[email protected]][(none)]> select * from stats_mysql_query_digest;
+-----------+--------------------+----------+--------------------+----------------------------------+------------+------------+------------+----------+----------+----------+
| hostgroup | schemaname         | username | digest             | digest_text                      | count_star | first_seen | last_seen  | sum_time | min_time | max_time |
+-----------+--------------------+----------+--------------------+----------------------------------+------------+------------+------------+----------+----------+----------+
| 100       | yayun              | yayun    | 0xA9518ABEA63705E6 | create table t1 ( id int)        | 1          | 1497577980 | 1497577980 | 79733    | 79733    | 79733    |
| 100       | yayun              | yayun    | 0x3765930C7143F468 | select * from t1                 | 1          | 1497577997 | 1497577997 | 1537     | 1537     | 1537     |
| 100       | yayun              | yayun    | 0x4BBB5CD4BC2CFD94 | insert into t1 select ?          | 1          | 1497577986 | 1497577986 | 33350    | 33350    | 33350    |
| 100       | information_schema | yayun    | 0x620B328FE9D6D71A | SELECT DATABASE()                | 1          | 1497577955 | 1497577955 | 4994     | 4994     | 4994     |
| 100       | information_schema | yayun    | 0x594F2C744B698066 | select USER()                    | 1          | 1497577951 | 1497577951 | 0        | 0        | 0        |
| 100       | information_schema | yayun    | 0x226CD90D52A2BA0B | select @@version_comment limit ? | 1          | 1497577951 | 1497577951 | 0        | 0        | 0        |
+-----------+--------------------+----------+--------------------+----------------------------------+------------+------------+------------+----------+----------+----------+
6 rows in set (0.00 sec)

[[email protected]][(none)]>

9. 可以看到读写都发送到了组100上面，组100是主库，说明没有读写分离。那是因为还有配置没有完成，我们需要自己定义规则。
定义路由规则，如：除select * from tb for update的select全部发送到slave，其他的的语句发送到master。

[[email protected]][(none)]> INSERT INTO mysql_query_rules(active,match_pattern,destination_hostgroup,apply) VALUES(1,'^SELECT.*FOR UPDATE$',100,1);
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]> INSERT INTO mysql_query_rules(active,match_pattern,destination_hostgroup,apply) VALUES(1,'^SELECT',1000,1);
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]> load mysql query rules to runtime;
Query OK, 0 rows affected (0.00 sec)

[[email protected]][(none)]> save mysql query rules to disk;
Query OK, 0 rows affected (0.02 sec)

[[email protected]][(none)]> select rule_id,active,match_pattern,destination_hostgroup,apply from runtime_mysql_query_rules;
+---------+--------+----------------------+-----------------------+-------+
| rule_id | active | match_pattern        | destination_hostgroup | apply |
+---------+--------+----------------------+-----------------------+-------+
| 1       | 1      | ^SELECT.*FOR UPDATE$ | 100                   | 1     |
| 2       | 1      | ^SELECT              | 1000                  | 1     |
+---------+--------+----------------------+-----------------------+-------+
2 rows in set (0.00 sec)

[[email protected]][(none)]>

清理掉统计信息，再次进行测试。

select * from stats_mysql_query_digest_reset;

再次运行读写，然后再查看，发现已经实现读写分离。

[[email protected]][(none)]> select * from stats_mysql_query_digest;                                                                    
+-----------+------------+----------+--------------------+-------------------------+------------+------------+------------+----------+----------+----------+
| hostgroup | schemaname | username | digest             | digest_text             | count_star | first_seen | last_seen  | sum_time | min_time | max_time |
+-----------+------------+----------+--------------------+-------------------------+------------+------------+------------+----------+----------+----------+
| 1000      | yayun      | yayun    | 0x3765930C7143F468 | select * from t1        | 1          | 1497578494 | 1497578494 | 21751    | 21751    | 21751    |
| 100       | yayun      | yayun    | 0x4BBB5CD4BC2CFD94 | insert into t1 select ? | 1          | 1497578492 | 1497578492 | 54852    | 54852    | 54852    |
+-----------+------------+----------+--------------------+-------------------------+------------+------------+------------+----------+----------+----------+
2 rows in set (0.00 sec)

[[email protected]][(none)]>

重点来了，如何配合MHA实现高可用呢？那么需要利用到proxysql里面的mysql_replication_hostgroups表。mysql_replication_hostgroups 表的主要作用是监视指定主机组中所有服务器的read_only值，并且根据read_only的值将服务器分配给写入器或读取器主机组，定义 hostgroup 的主从关系。ProxySQL monitor 模块会监控 HG 后端所有servers 的 read_only 变量，如果发现从库的 read_only 变为0、主库变为1，则认为角色互换了，自动改写 mysql_servers 表里面 hostgroup 关系，达到自动 Failover 效果。

我们看看mysql_replication_hostgroups表结构：

[[email protected]][(none)]> show create table mysql_replication_hostgroups\G
*************************** 1. row ***************************
       table: mysql_replication_hostgroups
Create Table: CREATE TABLE mysql_replication_hostgroups (
    writer_hostgroup INT CHECK (writer_hostgroup>=0) NOT NULL PRIMARY KEY,
    reader_hostgroup INT NOT NULL CHECK (reader_hostgroup<>writer_hostgroup AND reader_hostgroup>0),
    comment VARCHAR,
    UNIQUE (reader_hostgroup))
1 row in set (0.00 sec)

[[email protected]][(none)]>

3个字段很明显，写主机组，读主机组，备注。那么我们现在插入数据。我们的100是写，1000是读。

[[email protected]][(none)]> insert into  mysql_replication_hostgroups (writer_hostgroup,reader_hostgroup,comment)values(100,1000,'测试我的读写分离高可用');
Query OK, 1 row affected (0.00 sec)

[[email protected]][(none)]> load mysql servers to runtime;
Query OK, 0 rows affected (0.01 sec)

[[email protected]][(none)]> save mysql servers to disk;
Query OK, 0 rows affected (0.03 sec)

[[email protected]][(none)]> select * from runtime_mysql_replication_hostgroups;
+------------------+------------------+-----------------------------------+
| writer_hostgroup | reader_hostgroup | comment                           |
+------------------+------------------+-----------------------------------+
| 100              | 1000             | 测试我的读写分离高可用            |
+------------------+------------------+-----------------------------------+
1 row in set (0.00 sec)

[[email protected]][(none)]>

我们用vc-mysql-sniffer在从库抓一下。看看能看见什么。

[root@db_server_yayun_03 ~]# ./vc-mysql-sniffer 
# Time: 061617 10:26:45
# User@Host: unknown_user[unknown_user] @ 192.168.0.10:36371 []
# Query_time: 0.005198
SHOW GLOBAL VARIABLES LIKE 'read_only';
# Time: 061617 10:26:46
# User@Host: unknown_user[unknown_user] @ 192.168.0.10:36372 []
# Query_time: 0.000564
SHOW SLAVE STATUS;
# Time: 061617 10:26:46
# User@Host: unknown_user[unknown_user] @ 192.168.0.10:36372 []
# Query_time: 0.004008
SHOW GLOBAL VARIABLES LIKE 'read_only';
# Time: 061617 10:26:48
# User@Host: unknown_user[unknown_user] @ 192.168.0.10:36371 []
# Query_time: 0.004923
SHOW GLOBAL VARIABLES LIKE 'read_only';
# Time: 061617 10:26:49
# User@Host: unknown_user[unknown_user] @ 192.168.0.10:36372 []
# Query_time: 0.003617
SHOW GLOBAL VARIABLES LIKE 'read_only';

可以看见在检查read_only变量。我们先看看现在主机组的关系：

[[email protected]][(none)]> select * from runtime_mysql_servers;
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
| hostgroup_id | hostname     | port | status | weight | compression | max_connections | max_replication_lag | use_ssl | max_latency_ms | comment          |
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
| 100          | 192.168.0.20 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
| 1000         | 192.168.0.30 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
| 1000         | 192.168.0.40 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
3 rows in set (0.00 sec)

[[email protected]][(none)]>

可以看见100主机组是主库，1000主机组是从库。下面使用MHA在线变换主从关系。我们把从库40提升为主库。

[root@db_server_yayun_04 ~]# masterha_master_switch --master_state=alive --conf=/data/mha/3306/mha.cnf --new_master_host=192.168.0.40 --new_master_port=3306 --orig_master_is_new_slave 
Fri Jun 16 10:30:39 2017 - [info] MHA::MasterRotate version 0.57.
Fri Jun 16 10:30:39 2017 - [info] Starting online master switch..
Fri Jun 16 10:30:39 2017 - [info] 
Fri Jun 16 10:30:39 2017 - [info] * Phase 1: Configuration Check Phase..
Fri Jun 16 10:30:39 2017 - [info] 
Fri Jun 16 10:30:39 2017 - [warning] Global configuration file /etc/masterha_default.cnf not found. Skipping.
Fri Jun 16 10:30:39 2017 - [info] Reading application default configuration from /data/mha/3306/mha.cnf..
Fri Jun 16 10:30:39 2017 - [info] Updating application default configuration from /usr/local/bin/load_cnf..
Fri Jun 16 10:30:39 2017 - [info] Reading server configuration from /data/mha/3306/mha.cnf..
Fri Jun 16 10:30:39 2017 - [info] Setting max_ping_errors to 10, ping_interval to 3.
Fri Jun 16 10:30:39 2017 - [info] GTID failover mode = 1
Fri Jun 16 10:30:39 2017 - [info] Current Alive Master: 192.168.0.20(192.168.0.20:3306)
Fri Jun 16 10:30:39 2017 - [info] Alive Slaves:
Fri Jun 16 10:30:39 2017 - [info]   192.168.0.30(192.168.0.30:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:30:39 2017 - [info]     GTID ON
Fri Jun 16 10:30:39 2017 - [info]     Replicating from 192.168.0.20(192.168.0.20:3306)
Fri Jun 16 10:30:39 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Fri Jun 16 10:30:39 2017 - [info]   192.168.0.40(192.168.0.40:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:30:39 2017 - [info]     GTID ON
Fri Jun 16 10:30:39 2017 - [info]     Replicating from 192.168.0.20(192.168.0.20:3306)
Fri Jun 16 10:30:39 2017 - [info]     Primary candidate for the new Master (candidate_master is set)

It is better to execute FLUSH NO_WRITE_TO_BINLOG TABLES on the master before switching. Is it ok to execute on 192.168.0.20(192.168.0.20:3306)? (YES/no): y
Fri Jun 16 10:30:40 2017 - [info] Executing FLUSH NO_WRITE_TO_BINLOG TABLES. This may take long time..
Fri Jun 16 10:30:40 2017 - [info]  ok.
Fri Jun 16 10:30:40 2017 - [info] Checking MHA is not monitoring or doing failover..
Fri Jun 16 10:30:40 2017 - [info] Checking replication health on 192.168.0.30..
Fri Jun 16 10:30:40 2017 - [info]  ok.
Fri Jun 16 10:30:40 2017 - [info] Checking replication health on 192.168.0.40..
Fri Jun 16 10:30:40 2017 - [info]  ok.
Fri Jun 16 10:30:40 2017 - [info] 192.168.0.40 can be new master.
Fri Jun 16 10:30:40 2017 - [info] 
From:
192.168.0.20(192.168.0.20:3306) (current master)
 +--192.168.0.30(192.168.0.30:3306)
 +--192.168.0.40(192.168.0.40:3306)

To:
192.168.0.40(192.168.0.40:3306) (new master)
 +--192.168.0.30(192.168.0.30:3306)
 +--192.168.0.20(192.168.0.20:3306)

Starting master switch from 192.168.0.20(192.168.0.20:3306) to 192.168.0.40(192.168.0.40:3306)? (yes/NO): yes
Fri Jun 16 10:30:42 2017 - [info] Checking whether 192.168.0.40(192.168.0.40:3306) is ok for the new master..
Fri Jun 16 10:30:42 2017 - [info]  ok.
Fri Jun 16 10:30:42 2017 - [info] 192.168.0.20(192.168.0.20:3306): SHOW SLAVE STATUS returned empty result. To check replication filtering rules, temporarily executing CHANGE MASTER to a dummy host.
Fri Jun 16 10:30:42 2017 - [info] 192.168.0.20(192.168.0.20:3306): Resetting slave pointing to the dummy host.
Fri Jun 16 10:30:42 2017 - [info] ** Phase 1: Configuration Check Phase completed.
Fri Jun 16 10:30:42 2017 - [info] 
Fri Jun 16 10:30:42 2017 - [info] * Phase 2: Rejecting updates Phase..
Fri Jun 16 10:30:42 2017 - [info] 
Fri Jun 16 10:30:42 2017 - [info] Executing master ip online change script to disable write on the current master:
Fri Jun 16 10:30:42 2017 - [info]   /usr/local/bin/master_ip_online_change --command=stop --orig_master_host=192.168.0.20 --orig_master_ip=192.168.0.20 --orig_master_port=3306 --orig_master_user='root' --orig_master_password='123' --new_master_host=192.168.0.40 --new_master_ip=192.168.0.40 --new_master_port=3306 --new_master_user='root' --new_master_password='123' --orig_master_ssh_user=root --new_master_ssh_user=root   --orig_master_is_new_slave
Fri Jun 16 10:30:42 2017 783184 Set read_only on the new master.. ok.
Fri Jun 16 10:30:42 2017 786894 Drpping app user on the orig master..
Fri Jun 16 10:30:42 2017 787716 Waiting all running 3 threads are disconnected.. (max 1500 milliseconds)
{'Time' => '0','Command' => 'Sleep','db' => undef,'Id' => '282','Info' => undef,'User' => 'proxysql','State' => '','Host' => '192.168.0.10:54482'}
{'Time' => '1620','Command' => 'Binlog Dump GTID','db' => undef,'Id' => '296','Info' => undef,'User' => 'repl','State' => 'Master has sent all binlog to slave; waiting for more updates','Host' => '192.168.0.30:53451'}
{'Time' => '874','Command' => 'Binlog Dump GTID','db' => undef,'Id' => '310','Info' => undef,'User' => 'repl','State' => 'Master has sent all binlog to slave; waiting for more updates','Host' => '192.168.0.40:53136'}
Fri Jun 16 10:30:43 2017 294399 Waiting all running 2 threads are disconnected.. (max 1000 milliseconds)
{'Time' => '1621','Command' => 'Binlog Dump GTID','db' => undef,'Id' => '296','Info' => undef,'User' => 'repl','State' => 'Master has sent all binlog to slave; waiting for more updates','Host' => '192.168.0.30:53451'}
{'Time' => '875','Command' => 'Binlog Dump GTID','db' => undef,'Id' => '310','Info' => undef,'User' => 'repl','State' => 'Master has sent all binlog to slave; waiting for more updates','Host' => '192.168.0.40:53136'}
Fri Jun 16 10:30:43 2017 800400 Waiting all running 2 threads are disconnected.. (max 500 milliseconds)
{'Time' => '1621','Command' => 'Binlog Dump GTID','db' => undef,'Id' => '296','Info' => undef,'User' => 'repl','State' => 'Master has sent all binlog to slave; waiting for more updates','Host' => '192.168.0.30:53451'}
{'Time' => '875','Command' => 'Binlog Dump GTID','db' => undef,'Id' => '310','Info' => undef,'User' => 'repl','State' => 'Master has sent all binlog to slave; waiting for more updates','Host' => '192.168.0.40:53136'}
Fri Jun 16 10:30:44 2017 306937 Set read_only=1 on the orig master.. ok.
Fri Jun 16 10:30:44 2017 311939 Waiting all running 3 queries are disconnected.. (max 500 milliseconds)
{'Time' => '0','Command' => 'Sleep','db' => undef,'Id' => '281','Info' => undef,'User' => 'proxysql','State' => '','Host' => '192.168.0.10:54481'}
{'Time' => '1622','Command' => 'Binlog Dump GTID','db' => undef,'Id' => '296','Info' => undef,'User' => 'repl','State' => 'Master has sent all binlog to slave; waiting for more updates','Host' => '192.168.0.30:53451'}
{'Time' => '876','Command' => 'Binlog Dump GTID','db' => undef,'Id' => '310','Info' => undef,'User' => 'repl','State' => 'Master has sent all binlog to slave; waiting for more updates','Host' => '192.168.0.40:53136'}
Fri Jun 16 10:30:44 2017 810421 Killing all application threads..
Fri Jun 16 10:30:44 2017 816513 done.
SIOCSIFFLAGS: Cannot assign requested address
Fri Jun 16 10:30:45 2017 - [info]  ok.
Fri Jun 16 10:30:45 2017 - [info] Locking all tables on the orig master to reject updates from everybody (including root):
Fri Jun 16 10:30:45 2017 - [info] Executing FLUSH TABLES WITH READ LOCK..
Fri Jun 16 10:30:45 2017 - [info]  ok.
Fri Jun 16 10:30:45 2017 - [info] Orig master binlog:pos is mysql-bin.000001:1610.
Fri Jun 16 10:30:45 2017 - [info]  Waiting to execute all relay logs on 192.168.0.40(192.168.0.40:3306)..
Fri Jun 16 10:30:45 2017 - [info]  master_pos_wait(mysql-bin.000001:1610) completed on 192.168.0.40(192.168.0.40:3306). Executed 0 events.
Fri Jun 16 10:30:45 2017 - [info]   done.
Fri Jun 16 10:30:45 2017 - [info] Getting new master's binlog name and position..
Fri Jun 16 10:30:45 2017 - [info]  mysql-bin.000002:194
Fri Jun 16 10:30:45 2017 - [info]  All other slaves should start replication from here. Statement should be: CHANGE MASTER TO MASTER_HOST='192.168.0.40', MASTER_PORT=3306, MASTER_AUTO_POSITION=1, MASTER_USER='repl', MASTER_PASSWORD='xxx';
Fri Jun 16 10:30:45 2017 - [info] Executing master ip online change script to allow write on the new master:
Fri Jun 16 10:30:45 2017 - [info]   /usr/local/bin/master_ip_online_change --command=start --orig_master_host=192.168.0.20 --orig_master_ip=192.168.0.20 --orig_master_port=3306 --orig_master_user='root' --orig_master_password='123' --new_master_host=192.168.0.40 --new_master_ip=192.168.0.40 --new_master_port=3306 --new_master_user='root' --new_master_password='123' --orig_master_ssh_user=root --new_master_ssh_user=root   --orig_master_is_new_slave
Fri Jun 16 10:30:45 2017 744690 Set read_only=0 on the new master.
Fri Jun 16 10:30:45 2017 745124 Creating app user on the new master..
Fri Jun 16 10:30:45 2017 - [info]  ok.
Fri Jun 16 10:30:45 2017 - [info] 
Fri Jun 16 10:30:45 2017 - [info] * Switching slaves in parallel..
Fri Jun 16 10:30:45 2017 - [info] 
Fri Jun 16 10:30:45 2017 - [info] -- Slave switch on host 192.168.0.30(192.168.0.30:3306) started, pid: 9457
Fri Jun 16 10:30:45 2017 - [info] 
Fri Jun 16 10:30:45 2017 - [info] Log messages from 192.168.0.30 ...
Fri Jun 16 10:30:45 2017 - [info] 
Fri Jun 16 10:30:45 2017 - [info]  Waiting to execute all relay logs on 192.168.0.30(192.168.0.30:3306)..
Fri Jun 16 10:30:45 2017 - [info]  master_pos_wait(mysql-bin.000001:1610) completed on 192.168.0.30(192.168.0.30:3306). Executed 0 events.
Fri Jun 16 10:30:45 2017 - [info]   done.
Fri Jun 16 10:30:45 2017 - [info]  Resetting slave 192.168.0.30(192.168.0.30:3306) and starting replication from the new master 192.168.0.40(192.168.0.40:3306)..
Fri Jun 16 10:30:45 2017 - [info]  Executed CHANGE MASTER.
Fri Jun 16 10:30:45 2017 - [info]  Slave started.
Fri Jun 16 10:30:45 2017 - [info] End of log messages from 192.168.0.30 ...
Fri Jun 16 10:30:45 2017 - [info] 
Fri Jun 16 10:30:45 2017 - [info] -- Slave switch on host 192.168.0.30(192.168.0.30:3306) succeeded.
Fri Jun 16 10:30:45 2017 - [info] Unlocking all tables on the orig master:
Fri Jun 16 10:30:45 2017 - [info] Executing UNLOCK TABLES..
Fri Jun 16 10:30:45 2017 - [info]  ok.
Fri Jun 16 10:30:45 2017 - [info] Starting orig master as a new slave..
Fri Jun 16 10:30:45 2017 - [info]  Resetting slave 192.168.0.20(192.168.0.20:3306) and starting replication from the new master 192.168.0.40(192.168.0.40:3306)..
Fri Jun 16 10:30:45 2017 - [info]  Executed CHANGE MASTER.
Fri Jun 16 10:30:46 2017 - [info]  Slave started.
Fri Jun 16 10:30:46 2017 - [info] All new slave servers switched successfully.
Fri Jun 16 10:30:46 2017 - [info] 
Fri Jun 16 10:30:46 2017 - [info] * Phase 5: New master cleanup phase..
Fri Jun 16 10:30:46 2017 - [info] 
Fri Jun 16 10:30:46 2017 - [info]  192.168.0.40: Resetting slave info succeeded.
Fri Jun 16 10:30:46 2017 - [info] Switching master to 192.168.0.40(192.168.0.40:3306) completed successfully.

可以看见已经成功切换。我们再来看看运行主机组的状态：

[[email protected]][(none)]> select * from runtime_mysql_servers;
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
| hostgroup_id | hostname     | port | status | weight | compression | max_connections | max_replication_lag | use_ssl | max_latency_ms | comment          |
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
| 100          | 192.168.0.40 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
| 1000         | 192.168.0.30 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
| 1000         | 192.168.0.40 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
| 1000         | 192.168.0.20 | 3306 | ONLINE | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
+--------------+--------------+------+--------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
4 rows in set (0.00 sec)

[[email protected]][(none)]>

可以看见40自动设置成了主机组100，也就是master。我们进行读写测试。

[[email protected]][(none)]> select * from stats_mysql_query_digest;
+-----------+------------+----------+--------------------+-----------------------------+------------+------------+------------+----------+----------+----------+
| hostgroup | schemaname | username | digest             | digest_text                 | count_star | first_seen | last_seen  | sum_time | min_time | max_time |
+-----------+------------+----------+--------------------+-----------------------------+------------+------------+------------+----------+----------+----------+
| 1000      | yayun      | yayun    | 0x3765930C7143F468 | select * from t1            | 1          | 1497580568 | 1497580568 | 34920    | 34920    | 34920    |
| 100       | yayun      | yayun    | 0x5A680F86B3D8FB2B | select * from t1 for update | 1          | 1497580565 | 1497580565 | 9609     | 9609     | 9609     |
| 100       | yayun      | yayun    | 0x4BBB5CD4BC2CFD94 | insert into t1 select ?     | 1          | 1497580557 | 1497580557 | 133003   | 133003   | 133003   |
| 100       | yayun      | yayun    | 0x99531AEFF718C501 | show tables                 | 1          | 1497580544 | 1497580544 | 2051     | 2051     | 2051     |
| 100       | yayun      | yayun    | 0x74A739578E179F19 | show processlist            | 1          | 1497580532 | 1497580532 | 4335     | 4335     | 4335     |
+-----------+------------+----------+--------------------+-----------------------------+------------+------------+------------+----------+----------+----------+
5 rows in set (0.00 sec)

[[email protected]][(none)]>

可以看见除了for update语句和insert发送到主库，查询是发送到了从库。当然虽然1000主机组里面有40，也就是会有非常少量的查询才会发送到主库。这个没法避免。
下面进行主库宕机测试，启动mha监控，关闭主库。mha切换日志，可以看见30当选主库：

[root@db_server_yayun_04 log]# cat manager.log 
Fri Jun 16 10:38:59 2017 - [warning] Global configuration file /etc/masterha_default.cnf not found. Skipping.
Fri Jun 16 10:38:59 2017 - [info] Reading application default configuration from /data/mha/3306/mha.cnf..
Fri Jun 16 10:38:59 2017 - [info] Updating application default configuration from /usr/local/bin/load_cnf..
Fri Jun 16 10:38:59 2017 - [info] Reading server configuration from /data/mha/3306/mha.cnf..
Fri Jun 16 10:38:59 2017 - [info] Setting max_ping_errors to 10, ping_interval to 3.
Fri Jun 16 10:39:53 2017 - [warning] Global configuration file /etc/masterha_default.cnf not found. Skipping.
Fri Jun 16 10:39:53 2017 - [info] Reading application default configuration from /data/mha/3306/mha.cnf..
Fri Jun 16 10:39:53 2017 - [info] Updating application default configuration from /usr/local/bin/load_cnf..
Fri Jun 16 10:39:53 2017 - [info] Reading server configuration from /data/mha/3306/mha.cnf..
Fri Jun 16 10:39:53 2017 - [info] Setting max_ping_errors to 10, ping_interval to 3.
nfo]   192.168.0.20(192.168.0.20:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:38:59 2017 - [info]     GTID ON
Fri Jun 16 10:38:59 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:38:59 2017 - [info]   192.168.0.30(192.168.0.30:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:38:59 2017 - [info]     GTID ON
Fri Jun 16 10:38:59 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:38:59 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Fri Jun 16 10:38:59 2017 - [info] Current Alive Master: 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:38:59 2017 - [info] Checking slave configurations..
Fri Jun 16 10:38:59 2017 - [info] Checking replication filtering settings..
Fri Jun 16 10:38:59 2017 - [info]  binlog_do_db= , binlog_ignore_db= 
Fri Jun 16 10:38:59 2017 - [info]  Replication filtering check ok.
Fri Jun 16 10:38:59 2017 - [info] GTID (with auto-pos) is supported. Skipping all SSH and Node package checking.
Fri Jun 16 10:38:59 2017 - [info] Checking SSH publickey authentication settings on the current master..
Fri Jun 16 10:38:59 2017 - [info] HealthCheck: SSH to 192.168.0.40 is reachable.
Fri Jun 16 10:38:59 2017 - [info] 
192.168.0.40(192.168.0.40:3306) (current master)
 +--192.168.0.20(192.168.0.20:3306)
 +--192.168.0.30(192.168.0.30:3306)

Fri Jun 16 10:38:59 2017 - [warning] master_ip_failover_script is not defined.
Fri Jun 16 10:38:59 2017 - [warning] shutdown_script is not defined.
Fri Jun 16 10:38:59 2017 - [info] Set master ping interval 3 seconds.
Fri Jun 16 10:38:59 2017 - [warning] secondary_check_script is not defined. It is highly recommended setting it to check master reachability from two or more routes.
Fri Jun 16 10:38:59 2017 - [info] Starting ping health check on 192.168.0.40(192.168.0.40:3306)..
Fri Jun 16 10:38:59 2017 - [info] Ping(SELECT) succeeded, waiting until MySQL doesn't respond..
Fri Jun 16 10:39:26 2017 - [warning] Got error on MySQL select ping: 2006 (MySQL server has gone away)
Fri Jun 16 10:39:26 2017 - [info] Executing SSH check script: exit 0
Fri Jun 16 10:39:27 2017 - [info] HealthCheck: SSH to 192.168.0.40 is reachable.
Fri Jun 16 10:39:29 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:29 2017 - [warning] Connection failed 2 time(s)..
Fri Jun 16 10:39:32 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:32 2017 - [warning] Connection failed 3 time(s)..
Fri Jun 16 10:39:35 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:35 2017 - [warning] Connection failed 4 time(s)..
Fri Jun 16 10:39:38 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:38 2017 - [warning] Connection failed 5 time(s)..
Fri Jun 16 10:39:41 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:41 2017 - [warning] Connection failed 6 time(s)..
Fri Jun 16 10:39:44 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:44 2017 - [warning] Connection failed 7 time(s)..
Fri Jun 16 10:39:47 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:47 2017 - [warning] Connection failed 8 time(s)..
Fri Jun 16 10:39:50 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:50 2017 - [warning] Connection failed 9 time(s)..
Fri Jun 16 10:39:53 2017 - [warning] Got error on MySQL connect: 2013 (Lost connection to MySQL server at 'reading initial communication packet', system error: 111)
Fri Jun 16 10:39:53 2017 - [warning] Connection failed 10 time(s)..
Fri Jun 16 10:39:53 2017 - [warning] Master is not reachable from health checker!
Fri Jun 16 10:39:53 2017 - [warning] Master 192.168.0.40(192.168.0.40:3306) is not reachable!
Fri Jun 16 10:39:53 2017 - [warning] SSH is reachable.
Fri Jun 16 10:39:53 2017 - [info] Connecting to a master server failed. Reading configuration file /etc/masterha_default.cnf and /data/mha/3306/mha.cnf again, and trying to connect to all servers to check server status..
Fri Jun 16 10:39:53 2017 - [warning] Global configuration file /etc/masterha_default.cnf not found. Skipping.
Fri Jun 16 10:39:53 2017 - [info] Reading application default configuration from /data/mha/3306/mha.cnf..
Fri Jun 16 10:39:53 2017 - [info] Updating application default configuration from /usr/local/bin/load_cnf..
Fri Jun 16 10:39:53 2017 - [info] Reading server configuration from /data/mha/3306/mha.cnf..
Fri Jun 16 10:39:53 2017 - [info] Setting max_ping_errors to 10, ping_interval to 3.
Fri Jun 16 10:39:53 2017 - [info] GTID failover mode = 1
Fri Jun 16 10:39:53 2017 - [info] Dead Servers:
Fri Jun 16 10:39:53 2017 - [info]   192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:53 2017 - [info] Alive Servers:
Fri Jun 16 10:39:53 2017 - [info]   192.168.0.20(192.168.0.20:3306)
Fri Jun 16 10:39:53 2017 - [info]   192.168.0.30(192.168.0.30:3306)
Fri Jun 16 10:39:53 2017 - [info] Alive Slaves:
Fri Jun 16 10:39:53 2017 - [info]   192.168.0.20(192.168.0.20:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:53 2017 - [info]     GTID ON
Fri Jun 16 10:39:53 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:53 2017 - [info]   192.168.0.30(192.168.0.30:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:53 2017 - [info]     GTID ON
Fri Jun 16 10:39:53 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:53 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Fri Jun 16 10:39:53 2017 - [info] Checking slave configurations..
Fri Jun 16 10:39:53 2017 - [info] Checking replication filtering settings..
Fri Jun 16 10:39:53 2017 - [info]  Replication filtering check ok.
Fri Jun 16 10:39:53 2017 - [info] Master is down!
Fri Jun 16 10:39:53 2017 - [info] Terminating monitoring script.
Fri Jun 16 10:39:53 2017 - [info] Got exit code 20 (Master dead).
Fri Jun 16 10:39:53 2017 - [info] MHA::MasterFailover version 0.57.
Fri Jun 16 10:39:53 2017 - [info] Starting master failover.
Fri Jun 16 10:39:53 2017 - [info] 
Fri Jun 16 10:39:53 2017 - [info] * Phase 1: Configuration Check Phase..
Fri Jun 16 10:39:53 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] GTID failover mode = 1
Fri Jun 16 10:39:54 2017 - [info] Dead Servers:
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:54 2017 - [info] Checking master reachability via MySQL(double check)...
Fri Jun 16 10:39:54 2017 - [info]  ok.
Fri Jun 16 10:39:54 2017 - [info] Alive Servers:
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.20(192.168.0.20:3306)
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.30(192.168.0.30:3306)
Fri Jun 16 10:39:54 2017 - [info] Alive Slaves:
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.20(192.168.0.20:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:54 2017 - [info]     GTID ON
Fri Jun 16 10:39:54 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.30(192.168.0.30:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:54 2017 - [info]     GTID ON
Fri Jun 16 10:39:54 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:54 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Fri Jun 16 10:39:54 2017 - [info] Starting GTID based failover.
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] ** Phase 1: Configuration Check Phase completed.
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] * Phase 2: Dead Master Shutdown Phase..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] Forcing shutdown so that applications never connect to the current master..
Fri Jun 16 10:39:54 2017 - [warning] master_ip_failover_script is not set. Skipping invalidating dead master IP address.
Fri Jun 16 10:39:54 2017 - [warning] shutdown_script is not set. Skipping explicit shutting down of the dead master.
Fri Jun 16 10:39:54 2017 - [info] * Phase 2: Dead Master Shutdown Phase completed.
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] * Phase 3: Master Recovery Phase..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] * Phase 3.1: Getting Latest Slaves Phase..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] The latest binary log file/position on all slaves is mysql-bin.000002:497
Fri Jun 16 10:39:54 2017 - [info] Retrieved Gtid Set: 900f20f2-f48c-11e6-8d78-000c2930a8b9:1
Fri Jun 16 10:39:54 2017 - [info] Latest slaves (Slaves that received relay log files to the latest):
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.20(192.168.0.20:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:54 2017 - [info]     GTID ON
Fri Jun 16 10:39:54 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.30(192.168.0.30:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:54 2017 - [info]     GTID ON
Fri Jun 16 10:39:54 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:54 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Fri Jun 16 10:39:54 2017 - [info] The oldest binary log file/position on all slaves is mysql-bin.000002:497
Fri Jun 16 10:39:54 2017 - [info] Retrieved Gtid Set: 900f20f2-f48c-11e6-8d78-000c2930a8b9:1
Fri Jun 16 10:39:54 2017 - [info] Oldest slaves:
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.20(192.168.0.20:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:54 2017 - [info]     GTID ON
Fri Jun 16 10:39:54 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.30(192.168.0.30:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:54 2017 - [info]     GTID ON
Fri Jun 16 10:39:54 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:54 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] * Phase 3.3: Determining New Master Phase..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] Searching new master from slaves..
Fri Jun 16 10:39:54 2017 - [info]  Candidate masters from the configuration file:
Fri Jun 16 10:39:54 2017 - [info]   192.168.0.30(192.168.0.30:3306)  Version=5.7.17-log (oldest major version between slaves) log-bin:enabled
Fri Jun 16 10:39:54 2017 - [info]     GTID ON
Fri Jun 16 10:39:54 2017 - [info]     Replicating from 192.168.0.40(192.168.0.40:3306)
Fri Jun 16 10:39:54 2017 - [info]     Primary candidate for the new Master (candidate_master is set)
Fri Jun 16 10:39:54 2017 - [info]  Non-candidate masters:
Fri Jun 16 10:39:54 2017 - [info]  Searching from candidate_master slaves which have received the latest relay log events..
Fri Jun 16 10:39:54 2017 - [info] New master is 192.168.0.30(192.168.0.30:3306)
Fri Jun 16 10:39:54 2017 - [info] Starting master failover..
Fri Jun 16 10:39:54 2017 - [info] 
From:
192.168.0.40(192.168.0.40:3306) (current master)
 +--192.168.0.20(192.168.0.20:3306)
 +--192.168.0.30(192.168.0.30:3306)

To:
192.168.0.30(192.168.0.30:3306) (new master)
 +--192.168.0.20(192.168.0.20:3306)
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] * Phase 3.3: New Master Recovery Phase..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info]  Waiting all logs to be applied.. 
Fri Jun 16 10:39:54 2017 - [info]   done.
Fri Jun 16 10:39:54 2017 - [info]  Replicating from the latest slave 192.168.0.20(192.168.0.20:3306) and waiting to apply..
Fri Jun 16 10:39:54 2017 - [info]  Waiting all logs to be applied on the latest slave.. 
Fri Jun 16 10:39:54 2017 - [info]  Resetting slave 192.168.0.30(192.168.0.30:3306) and starting replication from the new master 192.168.0.20(192.168.0.20:3306)..
Fri Jun 16 10:39:54 2017 - [info]  Executed CHANGE MASTER.
Fri Jun 16 10:39:54 2017 - [info]  Slave started.
Fri Jun 16 10:39:54 2017 - [info]  Waiting to execute all relay logs on 192.168.0.30(192.168.0.30:3306)..
Fri Jun 16 10:39:54 2017 - [info]  master_pos_wait(mysql-bin.000001:1879) completed on 192.168.0.30(192.168.0.30:3306). Executed 4 events.
Fri Jun 16 10:39:54 2017 - [info]   done.
Fri Jun 16 10:39:54 2017 - [info]   done.
Fri Jun 16 10:39:54 2017 - [info] Getting new master's binlog name and position..
Fri Jun 16 10:39:54 2017 - [info]  mysql-bin.000002:459
Fri Jun 16 10:39:54 2017 - [info]  All other slaves should start replication from here. Statement should be: CHANGE MASTER TO MASTER_HOST='192.168.0.30', MASTER_PORT=3306, MASTER_AUTO_POSITION=1, MASTER_USER='repl', MASTER_PASSWORD='xxx';
Fri Jun 16 10:39:54 2017 - [info] Master Recovery succeeded. File:Pos:Exec_Gtid_Set: mysql-bin.000002, 459, 56195f28-36e2-11e7-991b-000c29e3f5ab:1-6,
900f20f2-f48c-11e6-8d78-000c2930a8b9:1
Fri Jun 16 10:39:54 2017 - [warning] master_ip_failover_script is not set. Skipping taking over new master IP address.
Fri Jun 16 10:39:54 2017 - [info] Setting read_only=0 on 192.168.0.30(192.168.0.30:3306)..
Fri Jun 16 10:39:54 2017 - [info]  ok.
Fri Jun 16 10:39:54 2017 - [info] ** Finished master recovery successfully.
Fri Jun 16 10:39:54 2017 - [info] * Phase 3: Master Recovery Phase completed.
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] * Phase 4: Slaves Recovery Phase..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] * Phase 4.1: Starting Slaves in parallel..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] -- Slave recovery on host 192.168.0.20(192.168.0.20:3306) started, pid: 9905. Check tmp log /data/mha/3306/log/192.168.0.20_3306_20170616103953.log if it takes time..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] Log messages from 192.168.0.20 ...
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info]  Resetting slave 192.168.0.20(192.168.0.20:3306) and starting replication from the new master 192.168.0.30(192.168.0.30:3306)..
Fri Jun 16 10:39:54 2017 - [info]  Executed CHANGE MASTER.
Fri Jun 16 10:39:54 2017 - [info]  Slave started.
Fri Jun 16 10:39:54 2017 - [info]  gtid_wait(56195f28-36e2-11e7-991b-000c29e3f5ab:1-6,
900f20f2-f48c-11e6-8d78-000c2930a8b9:1) completed on 192.168.0.20(192.168.0.20:3306). Executed 0 events.
Fri Jun 16 10:39:54 2017 - [info] End of log messages from 192.168.0.20.
Fri Jun 16 10:39:54 2017 - [info] -- Slave on host 192.168.0.20(192.168.0.20:3306) started.
Fri Jun 16 10:39:54 2017 - [info] All new slave servers recovered successfully.
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] * Phase 5: New master cleanup phase..
Fri Jun 16 10:39:54 2017 - [info] 
Fri Jun 16 10:39:54 2017 - [info] Resetting slave info on the new master..
Fri Jun 16 10:39:54 2017 - [info]  192.168.0.30: Resetting slave info succeeded.
Fri Jun 16 10:39:54 2017 - [info] Master failover to 192.168.0.30(192.168.0.30:3306) completed successfully.
Fri Jun 16 10:39:54 2017 - [info] 

----- Failover Report -----

mha: MySQL Master failover 192.168.0.40(192.168.0.40:3306) to 192.168.0.30(192.168.0.30:3306) succeeded

Master 192.168.0.40(192.168.0.40:3306) is down!

Check MHA Manager logs at db_server_yayun_04:/data/mha/3306/log/manager.log for details.

Started automated(non-interactive) failover.
Selected 192.168.0.30(192.168.0.30:3306) as a new master.
192.168.0.30(192.168.0.30:3306): OK: Applying all logs succeeded.
192.168.0.20(192.168.0.20:3306): OK: Slave started, replicating from 192.168.0.30(192.168.0.30:3306)
192.168.0.30(192.168.0.30:3306): Resetting slave info succeeded.
Master failover to 192.168.0.30(192.168.0.30:3306) completed successfully.

再来看看proxysql里面现在主机组的情况：

[[email protected]][(none)]> select * from runtime_mysql_servers;
+--------------+--------------+------+---------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
| hostgroup_id | hostname     | port | status  | weight | compression | max_connections | max_replication_lag | use_ssl | max_latency_ms | comment          |
+--------------+--------------+------+---------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
| 100          | 192.168.0.30 | 3306 | ONLINE  | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
| 1000         | 192.168.0.30 | 3306 | ONLINE  | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
| 1000         | 192.168.0.40 | 3306 | SHUNNED | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
| 1000         | 192.168.0.20 | 3306 | ONLINE  | 1      | 0           | 1000            | 10                  | 0       | 100            | test my proxysql |
+--------------+--------------+------+---------+--------+-------------+-----------------+---------------------+---------+----------------+------------------+
4 rows in set (0.01 sec)

[[email protected]][(none)]>

可以看见40已经是SHUNNED状态，这个已经自动被踢出。30当选主库，自动被设置在主机组100. 再次运行读写测试：

[[email protected]][(none)]> select * from stats_mysql_query_digest;
+-----------+--------------------+----------+--------------------+----------------------------------+------------+------------+------------+----------+----------+----------+
| hostgroup | schemaname         | username | digest             | digest_text                      | count_star | first_seen | last_seen  | sum_time | min_time | max_time |
+-----------+--------------------+----------+--------------------+----------------------------------+------------+------------+------------+----------+----------+----------+
| 100       | yayun              | yayun    | 0x4BBB5CD4BC2CFD94 | insert into t1 select ?          | 1          | 1497581038 | 1497581038 | 114743   | 114743   | 114743   |
| 1000      | information_schema | yayun    | 0x620B328FE9D6D71A | SELECT DATABASE()                | 1          | 1497581026 | 1497581026 | 31128    | 31128    | 31128    |
| 100       | information_schema | yayun    | 0x594F2C744B698066 | select USER()                    | 1          | 1497581025 | 1497581025 | 0        | 0        | 0        |
| 1000      | yayun              | yayun    | 0x3765930C7143F468 | select * from t1                 | 1          | 1497581045 | 1497581045 | 3283     | 3283     | 3283     |
| 100       | information_schema | yayun    | 0x226CD90D52A2BA0B | select @@version_comment limit ? | 1          | 1497581025 | 1497581025 | 0        | 0        | 0        |
+-----------+--------------------+----------+--------------------+----------------------------------+------------+------------+------------+----------+----------+----------+
5 rows in set (0.00 sec)

可以看见依然没有问题。到这里本文就结束了，上面是1主2从，如果是1主1从，那么从挂了怎么办呢？ 需要额外一条规则，那就是在mysql_servers的hostgroup 1000 里面要插一条主库的记录，然后把weight设小，当读不到从库，回去主库查询。

关于ProxySQL比较详细的文章可以参考如下：


ProxySQL之安装及配置详解

ProxySQL之读写分离与分库路由演示

 

作者：Atlas

出处：Atlas的博客  http://www.cnblogs.com/gomysql

您的支持是对博主最大的鼓励，感谢您的认真阅读。本文版权归作者所有，欢迎转载，但请保留该声明。如果您需要技术支持，本人亦提供有偿服务。


相关 [mha proxysql 分离] 推荐：

MHA+ProxySQL实现读写分离高可用 - yayun - 博客园

- -
最近在研究ProxySQL，觉得还挺不错的，所以就简单的折腾了一下，ProxySQL目前也是Percona在推荐的一个读写分离的中间件. 关于详细的介绍可以参考官方文档. 本文主要介绍的是MHA+ProxySQL读写分离以及高可用，ProxySQL的细节请参考文档，目前已经有人写的非常详细了，文章最后会给出链接.

ProxySQL+Mysql实现数据库读写分离实战

- - SegmentFault 最新的文章
前面也写过几篇关于Mysql数据的文章：. MySQL集群高可用架构之MHA. MySQL 同步复制及高可用方案总结. 官方工具｜MySQL Router 高可用原理与实战. 今天给大家带来的是关于数据库读写分离相关的实战操作. ProxySQL是一个高性能的MySQL中间件，拥有强大的规则引擎. 具有以下特性： http://www.proxysql.com/.

MySQL使用ProxySQL实现读写分离-毛虫小臭臭-51CTO博客

- -
ProxySQL是一个高性能的MySQL中间件，拥有强大的规则引擎. ProxySQL版本：proxysql-1.4.8-1-centos7.x86_64.rpm. Mysql版本：MySQL 5.7.22. ProxySQL主机IP：192.168.1.101. Mysql主库IP：192.168.1.102.

MHA高可用架构与Atlas读写分离 - 惨绿少年 - 博客园

- -
　　MHA（Master High Availability）目前在MySQL高可用方面是一个相对成熟的解决方案，它由日本DeNA公司youshimaton（现就职于Facebook公司）开发，是一套优秀的作为MySQL高可用性环境下故障切换和主从提升的高可用软件. 在MySQL故障切换过程中， MHA能做到在10~30秒之内自动完成数据库的故障切换操作，并且在进行故障切换的过程中，MHA能在最大程度上保证数据的一致性，以达到真正意义上的高可用.

MySQL HA Solution 2019(2)ProxySQL

- - 企业架构 - ITeye博客
| hostgroup_id | hostname      | port | gtid_port | status | weight | compression | max_connections | max_replication_lag | use_ssl | max_latency_ms | comment |.

mysql 高可用方案MHA介绍

- - CSDN博客数据库推荐文章
MySQL 大牛用Perl写的一套MySQL故障切换方案，来保证数据库系统的高可用.在宕机的时间内（通常10—30秒内），完成故障切换，部署MHA，可避免主从一致性问题，节约购买新服务器的费用，不影响服务器性能，易安装，不改变现有部署.    还支持在线切换，从当前运行master切换到一个新的master上面，只需要很短的时间（0.5-2秒内），此时仅仅阻塞写操作，并不影响读操作，便于主机硬件维护.

MySQL高可用性大杀器之MHA

- - OurMySQL
作为前提条件，应先配置MySQL复制，并设置 SSH公钥免密码登录. 下面以CentOS为例来说明，最好先安装 EPEL，不然YUM可能找不到某些软件包. MHA由Node和Manager组成，Node运行在每一台MySQL服务器上，也就是说，不管是MySQL主服务器，还是MySQL从服务器，都要安装Node，而Manager通常运行在独立的服务器上，但如果硬件资源吃紧，也可以用一台MySQL从服务器来兼职Manager的角色.

MySQL 5.5 keepalived + haproxy + MHA 自动容灾

- - CSDN博客数据库推荐文章
keepalived安装：. } } 在83上配置 vim /usr/local/etc/keepalived/keepalived.conf global_defs {. 在81和83上安装mha node，在82上安装mha manager. vi /usr/local/share/perl5/MHA/ManagerConst.pm 89 my $msg = $args{message}; 90 $msg = "" unless($msg); <=插入一行 91 chomp $msg;.

MySQL高可用方案MHA的部署和原理 - iVictor - 博客园

- -
MHA（Master High Availability）是一套相对成熟的MySQL高可用方案，能做到在0~30s内自动完成数据库的故障切换操作，在master服务器不宕机的情况下，基本能保证数据的一致性. 它由两部分组成：MHA Manager（管理节点）和MHA Node（数据节点）. 其中，MHA Manager可以单独部署在一台独立的机器上管理多个master-slave集群，也可以部署在一台slave上.