Shulou(Shulou.com)06/01 Report--

Official manual:

Https://mariadb.com/resources/blog/binlog-server

Reference article:

Http://www.linuxidc.com/Linux/2016-12/137892.htm

Http://www.sohu.com/a/120438391_487514

Cons: currently, binlog server does not support GTID replication.

Experimental topology diagram:

Step 1:

Create an account with replication permissions on Node1:

> grant replication client,replication slave,select on *. * to 'rpl'@ "192.168.2.%" identified by' rpl'

# this is the account for copying from the database, and it is also the account for maxscale to pull binlog. It has one more select permission than the regular slave account.

First, hang Node2 under node1 to simulate the architecture before binlog server:

The process is nothing more than exporting all the data of node1, restoring it on node2 and change master it to node1, skipping the specific steps.

Install maxscale on Node3:

Rpm-ivh maxscale-2.1.4-1.rhel.6.x86_64.rpm

Mkdir / data/binlog/-p

Chown maxscale.maxscale / data/binlog/-R

The vim/etc/maxscale.conf content is as follows:

[maxscale]

Threads=4 # set according to the number of CPU cores

# # Information about connecting to Master

[Replication]

Type=service

Router=binlogrouter

Version_string=5.6.36-log

# version_string parameter is used to pass the version information of the master library to the slave library, MaxScale sends server handshakepacket to clients

Router_options=server-id=13,heartbeat=30,transaction_safety=1,rcompatibility=1,send_slave_heartbeat=1

Binlogdir=/data/binlog # this directory belongs to the master group must be maxscale

User=rpl

Passwd=rpl

# description:

# server-id sets the id of maxscale and cannot have the same name as the master or slave database.

# heartbeat=30 means that when maxscale does not receive the binlog log pushed by the main database within 30 seconds, send a heartbeat check

# transaction_safety=1 is used to enable incomplete transaction detection in the binlog log. When MariaDBMaxScale starts, an error message may appear if the current binlog file is corrupted or an incomplete transaction is found. During normal operation, the binlog event is not assigned to the slave library until the transaction has been committed. The default value is off and transaction_safety= on is set to enable incomplete transaction detection. [similar to the role of relay_log_recovery = ON]

# send_slave_heartbeat=1 enables heartbeat check

# # Information provided to slave connection

[ReplicationListener]

Type=listener

Service=Replication

Protocol=MySQLClient

Port=5308

# # maxscale backend management port

[MaxAdmin Service]

Type=service

Router=cli

[MaxAdmin Listener]

Type=listener

Service=MaxAdmin Service

Protocol=maxscaled

Socket=default

Vim / data/binlog/master.ini adds the following:

[binlog_configuration]

Master_host=192.168.2.11 # address of the main library

Master_port=3306 # main library port number

Copy account of master_user=rpl # master

Copy password for master_password=rpl # master

Filestem=mysql # means that the pulled binlog file is named mysql.***. My main library is also named mysql.*.

Add this master.ini file so that after starting maxscale, you will automatically pull all the current binlog files of the main library (even if the binlog of the main library is automatically purge after it expires, the binlog on the maxscale server will still be saved)

Then, open the maxscale service on node3:

/ etc/init.d/maxscale start

Wait a moment, node3 will pull all the binlog of the main library.

The log is recorded in / var/log/maxscale/maxscale.log.

Ss-lnt | if port 5308 is up for grep.

Mysql-urpl-prpl-h 127.0.0.1-P 5308 can log in to the maxscale console, just like mysql.

Now let's add the new node4 from the library to node3's binlog server:

First, import the full backup data of node1 into node4

Then find something similar in the head-35 all.sql full backup:

A record like CHANGE MASTER TOMASTER_LOG_FILE='mysql.000004', MASTER_LOG_POS=2254.

Perform a change master operation on node4:

> CHANGE MASTER TO MASTER_HOST='192.168.2.13'

MASTER_PORT=5308

MASTER_USER='rpl'

MASTER_PASSWORD='rpl'

MASTER_LOG_FILE='mysql.000004'

MASTER_LOG_POS=2254

Note that in the change master operation above, we only changed the address and port of the master, the user name and password used for replication.

Because binlog server is actually the same as master's data, it only drags master's binlog directly.

To do the same, we can also hang node2 under binlog server.

On node2:

Show slave status\ G record Exec_Master_Log_Pos and Master_Log_File.

Stop slave

Reset slave all

Then use change master to point your superiors to binlog server.

Other maxscale commands:

On node3, execute show slave hosts; to see

You can also log in to the maxscale console:

Maxadmin-S / tmp/maxadmin.sock

You can use the help prompt for many other commands that view status, such as MaxScale > show services. That's not the point.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.