3. install and deploy Mgr cluster | explain Mgr in simple terms

> * GreatSQL The original content of the community shall not be used without authorization , For reprint, please contact the editor and indicate the source .

• 1. Installation preparation
• 2. initialization MySQL Server
• 3. initialization MGR First node
• 4. Continue setting up the other two nodes
• 5. towards MGR Write data to the cluster
• Reference material 、 file
• disclaimer

This article introduces how to use GreatSQL 8.0.25 Build a three node MGR colony .

1. Installation preparation

Prepare the following three servers ：

IP	port	role
172.16.16.10	3306	mgr1
172.16.16.11	3306	mgr2
172.16.16.12	3306	mgr3

Ensure that the network between the three nodes can be interconnected , And not targeted 3306 and 33061 Firewall interception rules for ports .

download GreatSQL Binary package , Download address ：https://gitee.com/GreatSQL/GreatSQL/releases .

This article takes CentOS x86_64 Environment, for example , The downloaded binary package name is ： GreatSQL-8.0.25-15-Linux-glibc2.28-x86_64.tar.xz, Put it in /usr/local Directory and unzip it ：

$ cd /usr/local
$ tar xf GreatSQL-8.0.25-15-Linux-glibc2.28-x86_64.tar.xz
$ cd GreatSQL-8.0.25-15-Linux-glibc2.28-x86_64
$ ls
bin    COPYING-jemalloc  include  LICENSE         LICENSE-test  mysqlrouter-log-rotate  README.router  run    support-files
cmake  docs              lib      LICENSE.router  man           README                  README-test    share  var

2. initialization MySQL Server

Get ready first /etc/my.cnf The configuration file ：

#/etc/my.cnf
[mysqld]
user = mysql
basedir=/usr/local/GreatSQL-8.0.25-15-Linux-glibc2.28-x86_64
datadir=/data/GreatSQL
port=3306
server_id=103306
log-bin
log_slave_updates=1
gtid_mode=ON
enforce_gtid_consistency=ON

This article can only be started normally MySQL Server And deployment MGR For the purpose , So this configuration file is extremely simple , If you want to use it on a formal occasion , You can refer to This profile .

Initialize first MySQL Server：

$ mkdir -p /data/GreatSQL && chown -R mysql:mysql /data/GreatSQL
$ /usr/local/GreatSQL-8.0.25-15-Linux-glibc2.28-x86_64/bin/mysqld --defaults-file=/etc/my.cnf --initialize-insecure

Be careful ： Don't use... In a production environment --initialize-insecure Option to initialize the installation , Because if you do this , Super administrator root By default, the account number is blank , Anyone can use this account to log in to the database , There are security risks , This is done for demonstration purposes only in this article .

start-up MySQL Server：

$ /usr/local/GreatSQL-8.0.25-15-Linux-glibc2.28-x86_64/bin/mysqld --defaults-file=/etc/my.cnf &

If nothing goes wrong , It can start normally MySQL Server. Use the same method to initialize the other two nodes .

Besides , Make a proposal to GreatSQL To join the system systemd In service , Easy to manage . For specific methods, please refer to this article ： take GreatSQL Add to system systemd service .

3. initialization MGR First node

Next, prepare to initialize MGR The first node of , Also known as Boot node .

modify /etc/my.cnf , Add the following lines and MGR Related configuration parameters ：

plugin_load_add='group_replication.so'
group_replication_group_name="aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaa1"
group_replication_local_address= "172.16.16.10:33061"
group_replication_group_seeds= "172.16.16.10:33061,172.16.16.11:33061,172.16.16.12:33061"
report-host=172.16.16.10

Options report-host The role of MGR Other nodes report the addresses used by this node , Avoid multiple host names on a server , May not be able to find the corresponding relationship correctly and make MGR Can't start problem . Besides , Set up report-host after , modify /etc/hosts It is not necessary to add the address and host name of each node to the system file .

in addition , Note that the port configured above is written as 33061 instead of 3306, This is for MGR The service specifies a dedicated communication port , The difference in MySQL Normal read / write service port . there 33061 The port number can be customized , For example, write 12345 It's fine too , Note that this port cannot be blocked by the firewall .

Use this configuration file , restart MySQL Server, After that, you should see that it has been loaded successfully group_replicaiton Plug in ：

mysql> show plugins;
...
+---------------------------------+----------+--------------------+----------------------+---------+
| Name                            | Status   | Type               | Library              | License |
+---------------------------------+----------+--------------------+----------------------+---------+
...
| group_replication               | ACTIVE   | GROUP REPLICATION  | group_replication.so | GPL     |
...

If not loaded correctly , You can also log in MySQL Server Load this manually plugin：

myqsl> install plugin group_replication soname 'group_replication.so';

Next , establish MGR Special account for services , And prepare the configuration MGR Service channel ：

# Each node should create a user separately , So there is no need to record this operation binlog And copy to other nodes 
mysql> set session sql_log_bin=0;
mysql> create user [email protected]'%' identified by 'repl';
mysql> GRANT BACKUP_ADMIN, REPLICATION SLAVE ON *.* TO `repl`@`%`;
# Continue to enable after the user is created binlog Record 
mysql> set session sql_log_bin=1;

# To configure MGR Service channel 
# Channel name  group_replication_recovery  Is constant , Do not modify 
mysql> CHANGE MASTER TO MASTER_USER='repl', MASTER_PASSWORD='repl' FOR CHANNEL 'group_replication_recovery';

Then execute the following command , Set it to MGR Boot node for （ Only the first node needs to do this ） Then you can start it directly MGR service ：

mysql> set global group_replication_bootstrap_group=ON;

mysql> start group_replication;

remind ： When the whole MGR When the cluster restarts , The first boot node should also be set to boot mode , Then start the other nodes . In addition , Do not set boot mode .

Then , see MGR Service status ：

mysql> select * from performance_schema.replication_group_members;
+---------------------------+--------------------------------------+--------------+-------------+--------------+-------------+----------------+
| CHANNEL_NAME              | MEMBER_ID                            | MEMBER_HOST  | MEMBER_PORT | MEMBER_STATE | MEMBER_ROLE | MEMBER_VERSION |
+---------------------------+--------------------------------------+--------------+-------------+--------------+-------------+----------------+
| group_replication_applier | 4ebd3504-11d9-11ec-8f92-70b5e873a570 | 172.16.16.10 |        3306 | ONLINE       | PRIMARY     | 8.0.25         |
+---------------------------+--------------------------------------+--------------+-------------+--------------+-------------+----------------+

Okay , The first node initialization is completed .

4. Continue setting up the other two nodes

Continue to use the following /etc/my.cnf Profile templates ：

#my.cnf
[mysqld]
user = mysql
basedir=/usr/local/GreatSQL-8.0.25-15-Linux-glibc2.28-x86_64
datadir=/data/GreatSQL
port=3306
server_id=113306
log-bin
log_slave_updates=1
gtid_mode=ON
enforce_gtid_consistency=ON

#mgr
plugin_load_add='group_replication.so'
group_replication_group_name="aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaa1"
group_replication_local_address= "172.16.16.11:33061"
group_replication_group_seeds= "172.16.16.10:33061,172.16.16.11:33061,172.16.16.12:33061"
report-host=172.16.16.11

remind ： Of the above options ,server_id、group_replication_local_address and report_host These three options should be modified to the correct values . In a MGR In the cluster , For each node server_id and server_uuid If only , however group_replication_group_name The value of is the same , It's time MGR Unique identity of the cluster .

restart MySQL Server After the instance （report-host Is a read-only option , It needs to be restarted to take effect ）, establish MGR Special service account and configuration MGR Service channel ：

mysql> set session sql_log_bin=0;
mysql> create user [email protected]'%' identified by 'repl';
mysql> GRANT BACKUP_ADMIN, REPLICATION SLAVE ON *.* TO `repl`@`%`;
mysql> set session sql_log_bin=1;

mysql> CHANGE MASTER TO MASTER_USER='repl', MASTER_PASSWORD='repl' FOR CHANNEL 'group_replication_recovery';

Then you can start directly MGR service （ Except for the first node , Other nodes do not need to set the boot mode ）：

mysql> start group_replication;

Look again MGR Node status ：

mysql> select * from performance_schema.replication_group_members;
+---------------------------+--------------------------------------+--------------+-------------+--------------+-------------+----------------+
| CHANNEL_NAME              | MEMBER_ID                            | MEMBER_HOST  | MEMBER_PORT | MEMBER_STATE | MEMBER_ROLE | MEMBER_VERSION |
+---------------------------+--------------------------------------+--------------+-------------+--------------+-------------+----------------+
| group_replication_applier | 4ebd3504-11d9-11ec-8f92-70b5e873a570 | 172.16.16.10 |        3306 | ONLINE       | PRIMARY     | 8.0.25         |
| group_replication_applier | 549b92bf-11d9-11ec-88e1-70b5e873a570 | 172.16.16.11 |        3306 | ONLINE       | SECONDARY   | 8.0.25         |
| group_replication_applier | 5596116c-11d9-11ec-8624-70b5e873a570 | 172.16.16.12 |        3306 | ONLINE       | SECONDARY   | 8.0.25         |
+---------------------------+--------------------------------------+--------------+-------------+--------------+-------------+----------------+

You can see that the cluster above has 3 Nodes are in ONLINE state , among 172.16.16.10 yes PRIMARY node , The other two are SECONDARY node , That is to say, the current cluster adopts Single owner Pattern . If multi master mode is adopted , Then the roles of all nodes are PRIMARY.

5. towards MGR Write data to the cluster

Next we connect to PRIMARY node , Create a test library table and write data ：

$mysql -h172.16.16.10 -uroot -Spath/mysql.sock
mysql> create database mgr;
mysql> use mgr;
mysql> create table t1(c1 int unsigned not null primary key);
mysql> insert into t1 select rand()*10240;
mysql> select * from t1;
+------+
| c1   |
+------+
| 8078 |
+------+

And then connect to one of them SECONDARY node , View just in PRIMARY Whether the written data can be seen ：

$mysql -h172.16.16.11 -uroot -Spath/mysql.sock
mysql> use mgr;
mysql> select * from t1;
+------+
| c1   |
+------+
| 8078 |
+------+

Confirm that the data can be read .

Come here , Three nodes are completed MGR Installation and deployment of cluster .

Reference material 、 file

• MySQL 8.0 Reference Manual
• Database kernel development - Wenzheng Lake
• Group Replication principle - Song libing

disclaimer

Due to limited personal level , There are inevitable mistakes and omissions in the column , Do not directly copy the commands in the document 、 The method is directly applied to the online production environment . Readers must fully understand and verify the test environment before formal implementation , Avoid damaging or damaging the production environment .