MySQL-MMM solution,

MySQL-MMM solution,

Reference:

This article briefly introduces the mmm solution and provides a simple verification.

I. MySQL-MMM solution 1. Introduction to MMM Solution

MMM (Multi-Master Replication Manager for MySQL) is a set of script suites for monitoring, failover, and management of MySQL Multi-Master Replication, and load balancing for read requests, it also provides a good read/write splitting architecture for MySQL.

Note:

2. Introduction to MMM Kit

The main functions of MMM kit are implemented through the following three scripts:

Ii. Verify the environment 1. Operating System

CentOS-6.7-x86_64

2. MySQL version

MySQL is 5.6.36: https://cdn.mysql.com//Downloads/MySQL-5.6/mysql-5.6.36.tar.gz

3. Topology

Role	Hosts	IP	Attributes	VIP	Description
Master1	Master	10.11.4.196	Write	10.11.4.191	Write Through VIP
Master2	Backup	10.11.4.197	Write | read	10.11.4.192 10.11.4.193	Master2 only provides read services when master1 is not faulty, while master1 takes over VIP
Slave	Slave	10.11.4.198	Read
Monitor	Mmm	10.11.4.199	Monitor

Iii. Configuration 1. Operating System Configuration hosts

# Four servers have the same configuration [root @ master ~] # Vim/etc/hosts10.11.4.196 master10.11.4.197 backup10.11.4.198 slave10.11.4.198 mmm

2. Create an account

User	Password	Privileges	Description
Mmm_monitor	Mmm_monitor	REPLICATION CLIENT	Monitor database status, including Master/Slave latency
Mmm_agent	Mmm_agent	SUPER, replication client, PROCESS	Modify the read_only status of the write server and redirect to the new master database.
Repl	Repl	REPLICATION SLAVE	Master/Slave replication user(Pre-configured)

# You only need to create an account on three db servers. The monitoring server does not need [root @ master ~] # Mysql-uroot-pEnter password: mysql> grant replication client on *. * TO 'mmm _ monitor' @ '10. 11.4.% 'identified BY 'mmm _ monitor'; mysql> grant super, replication client, process on *. * TO 'mmm _ agent' @ '10. 11.4.% 'identified BY 'mmm _ agent'; mysql> flush privileges;

3. Install MMM 1) 3 database servers

# The default yum source does not contain mmm. You need to install epel first. # You only need to install mysql-mmm-agnet [root @ master ~] on the database server. # Wget requests ~] # Rpm-ivh epel-release-6-8.noarch.rpm [root @ master ~] # Yum-y install mysql-mmm-agent

2) Monitoring Server

# The monitoring server can install all the mmm solution components. In fact, only the mysql-mmm-monitor component [root @ mmm ~] is started. # Wget requests ~] # Rpm-ivh Co., epel-release-6-8.noarch.rpm. [root @ mmm ~] # Yum-y install mysql-mmm *

3) MMM file path description

Path	Description
/Usr/libexec/mysql-mmm/	Script path
/Usr/share/perl5/vendor_perl/MMM/	Perl path of MMM
/Usr/sbin/	Executable command
/Etc/init. d/	Start the service
/Etc/mysql-mmm/	Configuration File
/Var/log/mysql-mmm/	Log Files

4. Configuration File 1) mmm_common.conf File

# Mmm_common.conf file, which must be configured on both the monitoring server and the database server, and the configuration is consistent; # The following is the configuration as planned [root @ mmm ~] # Vim/etc/mysql-mmm/mmm_common.conf # indicates the active master role. The read_only parameter must be enabled for all db servers. The Monitoring Agent automatically closes the read_only attribute of the writer server. active_master_role writer 2) mmm_mon.conf File
# Mmm_mon.conf file, Monitoring Server File configuration [root @ mmm ~] # Vim/etc/mysql-mmm/mmm_mon.confinclude mmm_common.conf <monitor> ip 127.0.0.1 pid_path/var/run/mysql-mmm/mmm_mond.pid bin_path/usr/libexec/mysql-mmm # cluster Status File, the original data status_path/var/lib/mysql-mmm/mmm_mond.status of the mmm_control show operation # The monitored db Server IP address ping_ips 10.11.4.196, 10.11.4.197, 10.11.4.198 # The default value is 60 s; # After the fault is restored, nodes with more than 60 s of offline will remain in the AWAITING_RECOVERY state. You need to manually set_online; # nodes with less than 60 s of offline and not in the flapping state, automatically online auto_set_online 20 # The kill_host_bin does not exist by default, though the monitor will # throw a warning about it missing. see the section 5.10 "Kill Host # Functionality" in the PDF documentation. # kill_host_bin/usr/libexec/mysql-mmm/monitor/kill_host # </monitor> 3) mmm_agent.conf File
# Mmm_agent.conf file, configure the proxy file on the three db servers [root @ master ~] # Vim/etc/mysql-mmm/mmm_agent.confinclude mmm_common.conf # The parameter after this is the host name of the current server; # modify the host Name of the three db servers according to the host name, this master is not required for the monitoring server
5. Start the process 1) monitoring/Management
# The monitoring server is not set when it is started by default; # If the configuration file is changed, the monitoring server or the db server must restart the process [root @ mmm ~] # Chkconfig -- level 35 mysql-mmm-monitor on [root @ mmm ~] # Service mysql-mmm-monitor start
2) proxy
# Operations on three db servers are similar to [root @ master ~] # Chkconfig -- level 35 mysql-mmm-agent on [root @ master ~] # Service mysql-mmm-agent start
6. iptables
# The mmm_agent service opens the local tcp9989 port for access by the monitoring/management end. The firewall is required to allow access. # The operations on the three db servers are similar to [root @ master ~]. # Vim/etc/sysconfig/iptables-a input-m state -- state NEW-m tcp-p tcp -- dport 9989-j ACCEPT [root @ master ~] # Service iptables restart
Iv. Verification 1. cluster node status
# Check whether the ping, mysql, and replication threads are normal [root @ mmm ~] # Mmm_control checks all
2. Cluster role status
# Each role has obtained the vip [root @ mmm ~] # Mmm_control show
3. Perform offline/online operations on the node. 1) offline
# The status of the backup node is "ADMIN_OFFLINE"; # The slave node obtains two read VIPs at the same time [root @ mmm ~] # Mmm_control set_offline backup [root @ mmm ~] # Mmm_control show
2) online
# Before the status of the backup node is restored to "ONLINE", there is a short "REPLICATION_FAIL" status [root @ mmm ~] # Mmm_control set_online backup [root @ mmm ~] # Mmm_control show
4. Execute the write switch operation. 1) view the current master of slave.
# The current master node is the master node [root @ slave ~] # Mysql-uroot-pxxxxxx-e 'show slave status \ G ;'
2) write Switch
# The write role has been switched to the backup node [root @ mmm ~] # Mmm_control move_role writer backup [root @ mmm ~] # Mmm_control show
 
3) check whether the master corresponding to the slave is switched
# The current master is the backup node [root @ slave ~] # Mysql-uroot-pXxxxxx-E 'show slave status \ G ;'
5. Summary 1. Logs
Monitoring end:/var/log/mysql-mmm/mmm_mond.log
AGENT:/var/log/mysql-mmm/mmm_agentd.log
2. Commands
In addition to common mmm_control commands, the mmm solution also provides the following commands:
Mmm_backup: Backup File
Mmm_restore: restore a file
Mmm_clone: clone a file
3. Notes