Fabric结合GTID实现Mysql复制高可用
mysql版本：Percona server 5.6.19-67.0 Percona Server (GPL), Release 67.0, Revision 618
Mysql安装参考：http://my.oschina.net/anthonyyau/blog/284092
Fabric State store安装参考：http://my.oschina.net/anthonyyau/blog/307165
半同步复制参考：http://my.oschina.net/anthonyyau/blog/269800
环境：
4个服务器实例运行Percona server 5.6.19-67.0，fabric node安装在单独的服务器，每个Mysql实例在一台服务器上；
fabric为启用复制的3个mysql实例提供高可用，应用使用fabric-aware的连接器路由事务和SQL语句到合适的服务器，透明的进行读和写操作；
当前仅仅支持异步primary backup复制(半同步复制需要手工完成)。primary处理所有写操作，Secondaries使用mysql复制从primary同步，可以进行读操作。
服务器信息：
一、mysql实例和fabric state store准备
Mysql安装好后，每个fabric管理的实例需要开启gtid、二进制日志(不需要启动复制，由fabric完成，但是不支持配置异步复制)，将以下配置放到[mysqld]段下面：
log-bin=mysql-bin
binlog_format=ROW
server-id = 29 #保证每个mysql实例唯一
log-slave-updates=true
gtid-mode=on
enforce-gtid-consistency=true
sync-master-info=1
数据库用户账号准备：
1、超级管理员账号，本例使用root@'172.17.42.1'，密码为admin@123；
2、fabric管理mysql实例的账号，本例使用fabric@'172.17.42.1'，密码为fabric@456；
3、mysql复制专用账号，本例使用fabric@'172.17.0.%'，密码为fabric@456，需要与fabric管理mysql的账号密码一致；
查看mysqlfabric命令帮助：
mysqlfabric help：显示简短的语法信息和帮忙命令
mysqlfabric help commands: 列出所有可用命令和描述
mysqlfabric help groups: 列出可用命令组
mysqlfabric help [group] [command]: 提供命令的详细帮助信息
二、使用Fabric创建mysql复制高可用组
1、创建组
# mysqlfabric group create my_group
Password for admin:
Procedure :
{ uuid = 3f7e82bc-4291-4002-8688-1929fc63ed3e,
finished = True,
success = True,
return = True,
activities =
}
要输入xml-rpc密码，可以将密码指定到fabric的配置文件，或者设置disable_authentication = yes，需要重启fabric(先mysqlfabric manage stop，然后修改配置文件，不然将报"Permission denied."错误)。
2、添加mysql实例到组
# mysqlfabric group add my_group 172.17.0.50:3306
Procedure :
{ uuid = 6e69a5a7-667b-4e63-92c3-2f9f4269d633,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric group add my_group 172.17.0.47:3306
Procedure :
{ uuid = 70a66b03-a16d-426d-a711-8f13da78fc8d,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric group add my_group 172.17.0.48:3306
Procedure :
{ uuid = 15a2e1ad-b726-4fa8-bdf3-a1701e870166,
finished = True,
success = True,
return = True,
activities =
}
添加实例到组错误：
# mysqlfabric --param=servers.user=fabric --param=servers.password=fabric@456 group add my_group 172.17.0.50:3306
Password for admin:
Procedure :
{ uuid = d2c0d969-5b8a-40c2-ba7a-1bc963c08824,
finished = True,
success = False,
return = ServerError: Error accessing server (172.17.0.50:3306).,
activities =
}
日志：
[DEBUG] 1409151193.274349 - Executor-4 - Statement (BEGIN, Params(()).
[DEBUG] 1409151193.274764 - Executor-4 - Executing _add_server
[DEBUG] 1409151193.274849 - Executor-4 - Statement (SELECT group_id, description, master_uuid, master_defined, status FROM groups WHERE group_id = %s, Params(('my_group',)).
[DEBUG] 1409151193.275494 - Executor-4 - Start executing function: discover_uuid((), {'connection_timeout': 5, 'address': '172.17.0.50:3306'}).
[DEBUG] 1409151193.276136 - Executor-4 - Error executing function: discover_uuid.
[DEBUG] 1409151193.276203 - Executor-4 - _add_server failed, executing compensation
[DEBUG] 1409151193.276255 - Executor-4 - Error accessing server (172.17.0.50:3306).
[DEBUG] 1409151193.276308 - Executor-4 - Statement (ROLLBACK, Params(()).
[DEBUG] 1409151193.276664 - Executor-4 - Complete job (60b0fa1f-3a39-43f3-aa71-46ea732aae84, 946e0199-53ca-4fdc-9b49-f134d44db476, mysql.fabric.services.server._add_server, Error).
将密码写到fabric配置文件是可以，使用命令行参数不能覆盖配置文件参数，暂时没有找到是什么原因。
3、查看组中信息
可以看到所有实例的状态都是SECONDARY
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': '7bd52611-2d44-11e4-af5f-3ecad7c2f82a', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.47:3306'}, {'status': 'SECONDARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
4、查看组健康详细信息
# mysqlfabric group health my_group
Command :
{ success = True
return = {'19a37552-2d44-11e4-af5c-763d1493518d': {'status': 'SECONDARY', 'is_alive': True, 'threads': {'is_configured': False}}, '7bd52611-2d44-11e4-af5f-3ecad7c2f82a': {'status': 'SECONDARY', 'is_alive': True, 'threads': {'is_configured': False}}, 'ade3ee53-2d44-11e4-af60-de532998e8a6': {'status': 'SECONDARY', 'is_alive': True, 'threads': {'is_configured': False}}}
activities =
}
5、提升和降级master服务
创建高可用组后，fabric没有意识到任何复制拓扑。需要提升一个为primary，降级剩余的服务器自动为secondaries(slaves)。
查看命令帮助：
# mysqlfabric help group promote
group promote group_id [--slave_id=NONE] [--update_only] [--synchronous]
如果只是想更新state store，跳过复制配置，使用--update_only参数。
如果slave没有提供，将选择一个最好的候选者，候选者必须开启二进制日志，同时跟master属于同一个组，跟master延时小。进行故障切换操作，选择这个候选者，同时将其他slave指向到新的master，同时更新state store。
提升一个mysql实例为master：
# mysqlfabric group promote my_group
Procedure :
{ uuid = 21eb4d58-d7ec-41eb-a0e5-560eb5976272,
finished = True,
success = True,
return = True,
activities =
}
执行同样的命名将设置不同的服务器为primary，同时降级当前primary并选取一个新的。如果当前primary错误，执行同样的命令能手动触发选取一个新的primary。
一个标记位"faulty"状态的服务器不能提升为secondary或primary，需要先转换成"spare"状态。使用命令mysqlfabric server set_status <server-address> spare
如果直接从"faulty"转换成"secondary"将报错：
# mysqlfabric server set_status 3ecc746f-2e05-11e4-b448-560d7281695e secondary
Procedure :
{ uuid = 06d0acb1-ad01-483b-a1d4-e13c4b775fcd,
finished = True,
success = False,
return = ServerError: Cannot change server's (3ecc746f-2e05-11e4-b448-560d7281695e) status from (FAULTY) to (SECONDARY).,
activities =
}
查看组状态和验证是否复制正常：
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'PRIMARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': '7bd52611-2d44-11e4-af5f-3ecad7c2f82a', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.47:3306'}, {'status': 'SECONDARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
使用show slave status查看slave状态：
# mysql -ufabric -pfabric@456 -h172.17.0.47 -e "show slave status\G;"
*************************** 1. row ***************************
Slave_IO_State: Connecting to master
Master_Host: 172.17.0.50
Master_User: fabric
Master_Port: 3306
Connect_Retry: 60
Master_Log_File:
Read_Master_Log_Pos: 4
Relay_Log_File: d4f404f647b0-relay-bin.000001
Relay_Log_Pos: 4
Relay_Master_Log_File:
Slave_IO_Running: Connecting
Slave_SQL_Running: Yes
Replicate_Do_DB:
Replicate_Ignore_DB:
Replicate_Do_Table:
Replicate_Ignore_Table:
Replicate_Wild_Do_Table:
Replicate_Wild_Ignore_Table:
Last_Errno: 0
Last_Error:
Skip_Counter: 0
Exec_Master_Log_Pos: 0
Relay_Log_Space: 151
Until_Condition: None
Until_Log_File:
Until_Log_Pos: 0
Master_SSL_Allowed: No
Master_SSL_CA_File:
Master_SSL_CA_Path:
Master_SSL_Cert:
Master_SSL_Cipher:
Master_SSL_Key:
Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
Last_IO_Errno: 1045
Last_IO_Error: error connecting to master '[email protected]:3306' - retry-time: 60 retries: 31
Last_SQL_Errno: 0
Last_SQL_Error:
Replicate_Ignore_Server_Ids:
Master_Server_Id: 0
Master_UUID:
Master_Info_File: /usr/local/Percona-Server-5.6.19-rel67.0-618.Linux.x86_64/data/master.info
SQL_Delay: 0
SQL_Remaining_Delay: NULL
Slave_SQL_Running_State: Slave has read all relay log; waiting for the slave I/O thread to update it
Master_Retry_Count: 86400
Master_Bind:
Last_IO_Error_Timestamp: 140827 16:42:50
Last_SQL_Error_Timestamp:
Master_SSL_Crl:
Master_SSL_Crlpath:
Retrieved_Gtid_Set:
Executed_Gtid_Set: 7bd52611-2d44-11e4-af5f-3ecad7c2f82a:1-6
Auto_Position: 1
看到连接错误：
因fabric使用其连接管理mysql实例的用户配置slave，所有需要重新授权slave访问master：
# mysql -uroot -padmin@123 -h172.17.0.50 -e "grant replication slave on *.* to fabric@'172.17.0.%' identified by 'fabric@456'"
# mysql -uroot -padmin@123 -h172.17.0.47 -e "grant replication slave on *.* to fabric@'172.17.0.%' identified by 'fabric@456'"
# mysql -uroot -padmin@123 -h172.17.0.48 -e "grant replication slave on *.* to fabric@'172.17.0.%' identified by 'fabric@456'"
授权后重新查看slave状态：看到连接正常，复制正常工作
# mysql -ufabric -pfabric@456 -h172.17.0.48 -e 'show slave status\G'
*************************** 1. row ***************************
Slave_IO_State: Waiting for master to send event
Master_Host: 172.17.0.50
Master_User: fabric
Master_Port: 3306
Connect_Retry: 60
Master_Log_File: mysql-bin.000004
Read_Master_Log_Pos: 3451
Relay_Log_File: e294ab366580-relay-bin.000003
Relay_Log_Pos: 448
Relay_Master_Log_File: mysql-bin.000004
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
Replicate_Do_DB:
Replicate_Ignore_DB:
Replicate_Do_Table:
Replicate_Ignore_Table:
Replicate_Wild_Do_Table:
Replicate_Wild_Ignore_Table:
Last_Errno: 0
Last_Error:
Skip_Counter: 0
Exec_Master_Log_Pos: 3451
Relay_Log_Space: 4169
Until_Condition: None
Until_Log_File:
Until_Log_Pos: 0
Master_SSL_Allowed: No
Master_SSL_CA_File:
Master_SSL_CA_Path:
Master_SSL_Cert:
Master_SSL_Cipher:
Master_SSL_Key:
Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
Last_IO_Errno: 0
Last_IO_Error:
Last_SQL_Errno: 0
Last_SQL_Error:
Replicate_Ignore_Server_Ids:
Master_Server_Id: 27
Master_UUID: 19a37552-2d44-11e4-af5c-763d1493518d
Master_Info_File: /usr/local/Percona-Server-5.6.19-rel67.0-618.Linux.x86_64/data/master.info
SQL_Delay: 0
SQL_Remaining_Delay: NULL
Slave_SQL_Running_State: Slave has read all relay log; waiting for the slave I/O thread to update it
Master_Retry_Count: 86400
Master_Bind:
Last_IO_Error_Timestamp:
Last_SQL_Error_Timestamp:
Master_SSL_Crl:
Master_SSL_Crlpath:
Retrieved_Gtid_Set: 19a37552-2d44-11e4-af5c-763d1493518d:1-10
Executed_Gtid_Set: 19a37552-2d44-11e4-af5c-763d1493518d:1-10
Auto_Position: 1
# mysql -ufabric -pfabric@456 -h172.17.0.47 -e 'show slave status\G'
*************************** 1. row ***************************
Slave_IO_State: Waiting for master to send event
Master_Host: 172.17.0.50
Master_User: fabric
Master_Port: 3306
Connect_Retry: 60
Master_Log_File: mysql-bin.000004
Read_Master_Log_Pos: 3451
Relay_Log_File: d4f404f647b0-relay-bin.000004
Relay_Log_Pos: 448
Relay_Master_Log_File: mysql-bin.000004
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
Replicate_Do_DB:
Replicate_Ignore_DB:
Replicate_Do_Table:
Replicate_Ignore_Table:
Replicate_Wild_Do_Table:
Replicate_Wild_Ignore_Table:
Last_Errno: 0
Last_Error:
Skip_Counter: 0
Exec_Master_Log_Pos: 3451
Relay_Log_Space: 1205
Until_Condition: None
Until_Log_File:
Until_Log_Pos: 0
Master_SSL_Allowed: No
Master_SSL_CA_File:
Master_SSL_CA_Path:
Master_SSL_Cert:
Master_SSL_Cipher:
Master_SSL_Key:
Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
Last_IO_Errno: 0
Last_IO_Error:
Last_SQL_Errno: 0
Last_SQL_Error:
Replicate_Ignore_Server_Ids:
Master_Server_Id: 27
Master_UUID: 19a37552-2d44-11e4-af5c-763d1493518d
Master_Info_File: /usr/local/Percona-Server-5.6.19-rel67.0-618.Linux.x86_64/data/master.info
SQL_Delay: 0
SQL_Remaining_Delay: NULL
Slave_SQL_Running_State: Slave has read all relay log; waiting for the slave I/O thread to update it
Master_Retry_Count: 86400
Master_Bind:
Last_IO_Error_Timestamp:
Last_SQL_Error_Timestamp:
Master_SSL_Crl:
Master_SSL_Crlpath:
Retrieved_Gtid_Set: 19a37552-2d44-11e4-af5c-763d1493518d:1-10
Executed_Gtid_Set: 19a37552-2d44-11e4-af5c-763d1493518d:1-10
Auto_Position: 1
6、启动或禁用故障检测
如果primary故障，可能想自动提升一个次级到primary，重定向剩余的secondaries到新的primary：
# mysqlfabric group activate my_group
Procedure :
{ uuid = 1d78324a-d39d-4098-8774-b0c2359a57ed,
finished = True,
success = True,
return = True,
activities =
}
状态检测语句：
[DEBUG] 1409157170.648943 - FailureDetector(my_group) - Statement (SELECT server_uuid, server_address, mode, status, weight FROM servers WHERE group_id = %s, Params(('my_group',)).
[DEBUG] 1409157172.655084 - FailureDetector(my_group) - Statement (SELECT group_id, description, master_uuid, master_defined, status FROM groups WHERE group_id = %s, Params(('my_group',)).
测试关闭当前master，看是否选取新的master，fabric能自动进程故障切换：
查看当前组的状态：当前master为"172.17.0.51"
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'PRIMARY', 'server_uuid': '3ecc746f-2e05-11e4-b448-560d7281695e', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.51:3306'}, {'status': 'SECONDARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
关闭当前master实例：
# mysqladmin -h172.17.0.51 -uroot -padmin@123 shutdown
关闭172.17.0.51实例后，fabric已经标记它的状态为"FAULTY"；
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'FAULTY', 'server_uuid': '3ecc746f-2e05-11e4-b448-560d7281695e', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.51:3306'}, {'status': 'PRIMARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
重新启动172.17.0.51:3306实例，但是状态不会恢复：
# mysql -h172.17.0.51 -uroot -padmin@123 -e "show master status"
+------------------+----------+--------------+------------------+-------------------------------------------+
| File | Position | Binlog_Do_DB | Binlog_Ignore_DB | Executed_Gtid_Set |
+------------------+----------+--------------+------------------+-------------------------------------------+
| mysql-bin.000003 | 191 | | | 19a37552-2d44-11e4-af5c-763d1493518d:1-10 |
+------------------+----------+--------------+------------------+-------------------------------------------+
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'FAULTY', 'server_uuid': '3ecc746f-2e05-11e4-b448-560d7281695e', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.51:3306'}, {'status': 'PRIMARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
需要手动设置服务器的状态：
查看帮助命令：
# mysqlfabric help server set_status
server set_status server_id status [--update_only] [--synchronous]
Set a server's status.
支持的状态值为primary,secondary,spare,faulty。
能手动改变服务器的状态，改变到spare，避免写和读访问，同时保证当故障切换时不会被选取。
改变错误状态的实例(172.17.0.51)的状态为secondary，注意需要先转换成spare：
# mysqlfabric server set_status 3ecc746f-2e05-11e4-b448-560d7281695e spare
Procedure :
{ uuid = e96802d7-5fc1-4a70-94d4-4d8d31d13960,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric server set_status 3ecc746f-2e05-11e4-b448-560d7281695e secondary
Procedure :
{ uuid = c74edd89-ac4a-45ce-8d18-22423ed73561,
finished = True,
success = True,
return = True,
activities =
}
查看状态：
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': '3ecc746f-2e05-11e4-b448-560d7281695e', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.51:3306'}, {'status': 'PRIMARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
禁用故障检测自动切换：
# mysqlfabric group deactivate my_group
Procedure :
{ uuid = a88d2bb7-f0c7-46d8-925a-c982287d6447,
finished = True,
success = True,
return = True,
activities =
}
7、执行更新和查询
应用程序通过fabric-aware连接器连接到fabric，获取服务器、组信息。
测试脚本：
# cat fabric_test1.py
#!/usr/bin/env python
#coding:utf-8
# Author: AnthonyYau --<[email protected]>
# Purpose: fabric test scripts1.
# Created: 2014/8/26
import sys
import mysql.connector
from mysql.connector import fabric
def main():
try:
conn = mysql.connector.connect(
fabric = {"host":"localhost","port":32274,"username":"admin",
"password":"admin"}, #指定使用xml-rpm连接fabric的用户名和密码
user = 'root',database = 'test',password = 'admin@123',autocommit = True #指定mysql实例的用户名和密码
)
except Exception,e:
print e
sys.exit(1)
else:
conn.set_property(mode=fabric.MODE_READWRITE,group="my_group") #指定访问模式，以及需要访问的HA组
cur = conn.cursor()
cur.execute(
"CREATE TABLE IF NOT EXISTS subscribers (" #创建表
" sub_no INT,"
" first_name CHAR(40),"
" last_name CHAR(4)"
")"
)
finally:
conn.close()
if __name__ == '__main__':
main()
# cat fabric_test2.py
#!/usr/bin/env python
#coding:utf-8
# Author: AnthonyYau --<[email protected]>
# Purpose: fabric test scripts2.
# Created: 2014/8/26
import sys
import mysql.connector
from mysql.connector import fabric
def add_subscriber(conn,sub_no,first_name,last_name):
conn.set_property(group="my_group",mode=fabric.MODE_READWRITE) #指定访问模式，以及访问的组
cur = conn.cursor()
cur.execute(
"INSERT INTO subscribers VALUES (%s,%s,%s)",
(sub_no,first_name,last_name)
)
def main():
try:
conn = mysql.connector.connect(
fabric = {"host":"localhost","port":32274,"username":"admin",
"password":"admin"}, #指定使用xml-rpc连接fabric的用户名和密码
user="root",database="test",password="admin@123",autocommit=True #指定访问mysql实例的用户名和密码
)
except Exception,e:
print e
sys.exit(1)
else:
#conn.set_property(group="my_group",mode=fabric.MODE_READWRITE)
add_subscriber(conn, 72, "Billy", "Fish") #调用函数，插入数据
add_subscriber(conn, 500, "Billy", "Joel")
add_subscriber(conn, 1500, "Arthur", "Askey")
add_subscriber(conn, 5000, "Billy", "Fish")
add_subscriber(conn, 15000, "Jimmy", "White")
add_subscriber(conn, 17542, "Bobby", "Ball")
finally:
conn.close()
if __name__ == '__main__':
执行测试程序：
python fabric_test1.py
python fabric_test2.py
检查结果(查看两台从服务器的数据)：当前slave为172.17.0.50和172.17.0.51
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': '3ecc746f-2e05-11e4-b448-560d7281695e', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.51:3306'}, {'status': 'PRIMARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
使用脚本检查结果：
# python fabric_test3.py
72 Billy Fish
500 Billy Joel
1500 Arthur Aske
5000 Billy Fish
15000 Jimmy Whit
17542 Bobby Ball
查询脚本如下：
#!/usr/bin/env python
#coding:utf-8
# Author: AnthonyYau --<[email protected]>
# Purpose: fabric test scripts3.
# Created: 2014/8/27
import sys
import mysql.connector
from mysql.connector import fabric
def main():
try:
conn = mysql.connector.connect(
fabric = {"host":"localhost","port":32274,"username":"admin",
"password":"admin"},
user="root",database="test",password="admin@123",autocommit=True
)
except Exception,e:
print e
sys.exit(1)
else:
conn.set_property(group="my_group",mode=fabric.MODE_READONLY) #使用readonly访问模式(从slave获取数据)，也可以使用readwrite
cur = conn.cursor()
cur.execute('select * from test.subscribers')
results = cur.fetchall()
if results:
for id,f_name,l_name in results:
print "%s\t%s\t%s" %(id,f_name,l_name)
finally:
conn.close()
if __name__ == '__main__':
main()
使用mysql客户端进行检查：
# mysql -uroot -padmin@123 -h172.17.0.51 -e "select * from test.subscribers"
+--------+------------+-----------+
| sub_no | first_name | last_name |
+--------+------------+-----------+
| 72 | Billy | Fish |
| 500 | Billy | Joel |
| 1500 | Arthur | Aske |
| 5000 | Billy | Fish |
| 15000 | Jimmy | Whit |
| 17542 | Bobby | Ball |
+--------+------------+-----------+
[root@yau620 ~]# mysql -uroot -padmin@123 -h172.17.0.50 -e "select * from test.subscribers"
+--------+------------+-----------+
| sub_no | first_name | last_name |
+--------+------------+-----------+
| 72 | Billy | Fish |
| 500 | Billy | Joel |
| 1500 | Arthur | Aske |
| 5000 | Billy | Fish |
| 15000 | Jimmy | Whit |
| 17542 | Bobby | Ball |
+--------+------------+-----------+
三、组维护
1、找出哪些服务器在组中，查看组状态
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'PRIMARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': '7bd52611-2d44-11e4-af5f-3ecad7c2f82a', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.47:3306'}, {'status': 'SECONDARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
2、从组中移除mysql实例
有时需要将secondaries离线进行维护，在停止服务之前，需要先从组中移除它，避免fabric错误检查触发任何动作：
# mysqlfabric help group remove
group remove group_id server_id [--synchronous]
Remove a server from a group.
如从my_group组中移除server_uuid为7bd52611-2d44-11e4-af5f-3ecad7c2f82a的实例：
# mysqlfabric group remove my_group 7bd52611-2d44-11e4-af5f-3ecad7c2f82a
Procedure :
{ uuid = 07317ba2-497c-43f0-b8a9-e43ef9ded834,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'PRIMARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
然后可以重新添加到组，将自动开启复制：
# mysqlfabric group add my_group 172.17.0.47:3306
Procedure :
{ uuid = 39344790-3af8-4049-92a6-669bd2642a1c,
finished = True,
success = True,
return = True,
activities =
}
查看复制状态：
# mysql -ufabric -pfabric@456 -h172.17.0.47 -e 'show slave status\G'
*************************** 1. row ***************************
Slave_IO_State: Waiting for master to send event
Master_Host: 172.17.0.50
Master_User: fabric
Master_Port: 3306
Connect_Retry: 60
Master_Log_File: mysql-bin.000004
Read_Master_Log_Pos: 3451
Relay_Log_File: d4f404f647b0-relay-bin.000002
Relay_Log_Pos: 408
Relay_Master_Log_File: mysql-bin.000004
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
Replicate_Do_DB:
Replicate_Ignore_DB:
Replicate_Do_Table:
Replicate_Ignore_Table:
Replicate_Wild_Do_Table:
Replicate_Wild_Ignore_Table:
Last_Errno: 0
Last_Error:
Skip_Counter: 0
Exec_Master_Log_Pos: 3451
Relay_Log_Space: 619
Until_Condition: None
Until_Log_File:
Until_Log_Pos: 0
Master_SSL_Allowed: No
Master_SSL_CA_File:
Master_SSL_CA_Path:
Master_SSL_Cert:
Master_SSL_Cipher:
Master_SSL_Key:
Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
Last_IO_Errno: 0
Last_IO_Error:
Last_SQL_Errno: 0
Last_SQL_Error:
Replicate_Ignore_Server_Ids:
Master_Server_Id: 27
Master_UUID: 19a37552-2d44-11e4-af5c-763d1493518d
Master_Info_File: /usr/local/Percona-Server-5.6.19-rel67.0-618.Linux.x86_64/data/master.info
SQL_Delay: 0
SQL_Remaining_Delay: NULL
Slave_SQL_Running_State: Slave has read all relay log; waiting for the slave I/O thread to update it
Master_Retry_Count: 86400
Master_Bind:
Last_IO_Error_Timestamp:
Last_SQL_Error_Timestamp:
Master_SSL_Crl:
Master_SSL_Crlpath:
Retrieved_Gtid_Set:
Executed_Gtid_Set: 19a37552-2d44-11e4-af5c-763d1493518d:1-10
Auto_Position: 1
3、添加一个全新的Mysql实例(实例IP为172.17.0.51)到HA group组
创建数据库复制账号：
# mysql -uroot -padmin@123 -h172.17.0.51 -e "grant replication slave on *.* to fabric@'172.17.0.%' identified by 'fabric@456'"
# mysql -uroot -padmin@123 -h172.17.0.51 -e "show slave status"
添加到my_group组：
# mysqlfabric group add my_group 172.17.0.51:3306
Procedure :
{ uuid = 45c32c66-6e60-44d8-8406-308de01b1ef9,
finished = True,
success = True,
return = True,
activities =
}
查看复制状态，fabric已经开启复制：
# mysql -uroot -padmin@123 -h172.17.0.51 -e "show slave status\G"
*************************** 1. row ***************************
Slave_IO_State: Waiting for master to send event
Master_Host: 172.17.0.50
Master_User: fabric
Master_Port: 3306
Connect_Retry: 60
Master_Log_File: mysql-bin.000004
Read_Master_Log_Pos: 3451
Relay_Log_File: 9355b5f279ad-relay-bin.000003
Relay_Log_Pos: 448
Relay_Master_Log_File: mysql-bin.000004
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
Replicate_Do_DB:
Replicate_Ignore_DB:
Replicate_Do_Table:
Replicate_Ignore_Table:
Replicate_Wild_Do_Table:
Replicate_Wild_Ignore_Table:
Last_Errno: 0
Last_Error:
Skip_Counter: 0
Exec_Master_Log_Pos: 3451
Relay_Log_Space: 4169
Until_Condition: None
Until_Log_File:
Until_Log_Pos: 0
Master_SSL_Allowed: No
Master_SSL_CA_File:
Master_SSL_CA_Path:
Master_SSL_Cert:
Master_SSL_Cipher:
Master_SSL_Key:
Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No
Last_IO_Errno: 0
Last_IO_Error:
Last_SQL_Errno: 0
Last_SQL_Error:
Replicate_Ignore_Server_Ids:
Master_Server_Id: 27
Master_UUID: 19a37552-2d44-11e4-af5c-763d1493518d
Master_Info_File: /usr/local/Percona-Server-5.6.19-rel67.0-618.Linux.x86_64/data/master.info
SQL_Delay: 0
SQL_Remaining_Delay: NULL
Slave_SQL_Running_State: Slave has read all relay log; waiting for the slave I/O thread to update it
Master_Retry_Count: 86400
Master_Bind:
Last_IO_Error_Timestamp:
Last_SQL_Error_Timestamp:
Master_SSL_Crl:
Master_SSL_Crlpath:
Retrieved_Gtid_Set: 19a37552-2d44-11e4-af5c-763d1493518d:1-10
Executed_Gtid_Set: 19a37552-2d44-11e4-af5c-763d1493518d:1-10
Auto_Position: 1
4、关闭primary
primary不能从组中移除，使用group demote关闭在组中的任何primary，不会重新选取一个新master，同时也不会关闭故障检测。
# mysqlfabric group demote my_group
Procedure :
{ uuid = 448df7ff-f0b5-4868-be15-8b505ec38c19,
finished = True,
success = True,
return = True,
activities =
}
查看组状态：所有mysql实例的状态都已经变成SECONDARY
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': '3ecc746f-2e05-11e4-b448-560d7281695e', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.51:3306'}, {'status': 'SECONDARY', 'server_uuid': '7bd52611-2d44-11e4-af5f-3ecad7c2f82a', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.47:3306'}, {'status': 'SECONDARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
5、手动提升一个服务组，选取一个master
当前master是172.17.0.48：
# mysqlfabric group promote my_group
Procedure :
{ uuid = d3072441-a023-42bd-8fe8-6746a620189d,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': '3ecc746f-2e05-11e4-b448-560d7281695e', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.51:3306'}, {'status': 'SECONDARY', 'server_uuid': '7bd52611-2d44-11e4-af5f-3ecad7c2f82a', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.47:3306'}, {'status': 'PRIMARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
重新选取一个新master：fabric已经自动选取了172.17.0.47作为master
# mysqlfabric group promote my_group
Procedure :
{ uuid = c779bfa6-5a57-4d60-9899-d64b85e9bee3,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric group lookup_servers my_group
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '19a37552-2d44-11e4-af5c-763d1493518d', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.50:3306'}, {'status': 'SECONDARY', 'server_uuid': '3ecc746f-2e05-11e4-b448-560d7281695e', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.51:3306'}, {'status': 'PRIMARY', 'server_uuid': '7bd52611-2d44-11e4-af5f-3ecad7c2f82a', 'mode': 'READ_WRITE', 'weight': 1.0, 'address': '172.17.0.47:3306'}, {'status': 'SECONDARY', 'server_uuid': 'ade3ee53-2d44-11e4-af60-de532998e8a6', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.48:3306'}]
activities =
}
6、移除组(使用group destroy)
如果组没有服务器，是空的，能移除组
# mysqlfabric group create my_test
Procedure :
{ uuid = d53ba76d-1ed0-4e9d-96b3-01102b46f0bf,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric group destroy my_test
Procedure :
{ uuid = 7046fea7-6e87-4dc0-afd4-e21ea399f9c1,
finished = True,
success = True,
return = True,
activities =
}
也能强制的移除非空的组(使用--force)：
# mysqlfabric group create my_test
Procedure :
{ uuid = be15e7bd-e597-4e67-aa4d-4ff34e0d3af8,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric group add my_test 172.17.0.52:3306
Procedure :
{ uuid = da981fe4-eeaf-4a3f-a770-d82d60a2ad64,
finished = True,
success = True,
return = True,
activities =
}
# mysqlfabric group lookup_servers my_test
Command :
{ success = True
return = [{'status': 'SECONDARY', 'server_uuid': '7bd52611-2d44-11e4-af5f-3ecad7c2f82a', 'mode': 'READ_ONLY', 'weight': 1.0, 'address': '172.17.0.52:3306'}]
activities =
}
# mysqlfabric group destroy my_test
Procedure :
{ uuid = 63b0ec92-ffa2-4f54-9c21-530b1532d21c,
finished = True,
success = False,
return = GroupError: Group (my_test) is not empty.,
activities =
}
# mysqlfabric group destroy my_test --force
Procedure :
{ uuid = 37c5b49b-c451-41ad-9193-c9e1f06edaa9,
finished = True,
success = True,
return = True,
activities =
}
四、总结
1、使用fabric实现mysql复制高可用，以组为单位对mysql实例进行管理，可以管理多个HA组；
2、当前fabric只能实现自动配置异步复制，不支持半同步复制；
3、mysql实例进行恢复后，fabric不会自动改变实例的状态，需要手工修改；
4、能在线添加和删除mysql实例；
5、fabric state store和fabric note当前是单点，虽然fabric-aware连接器可以缓存路由信息(可以配置ttl值)；
6、使用fabric不需要对slave配置vip，不像MHA和MMM等需要对vip进行管理；
7、mysql实例需要指出gtid，也是就要升级到mysql 5.6.10以上版本；
原文链接：
http://my.oschina.net/anthonyyau/blog/307355
参考：
1、http://www.clusterdb.com/mysql-fabric/mysql-fabric-adding-high-availability-to-mysql
2、官档：http://dev.mysql.com/doc/mysql-utilities/1.5/en/fabric-quick-start-replication.html