How to Build a High-Availability MySQL Cluster with MHA: Step‑by‑Step Guide

The previous article introduced MHA's function and principle; this guide shows how to build an MHA environment, simulate a master failure, and verify automatic failover.

Goal

Build an MHA environment, simulate a master failure, and confirm that a new master is elected correctly.

Setup Process

Initially four servers were planned: host1 as master, host2 and host3 as slaves, and host4 as the MHA manager. Due to limited resources the demo uses three servers.

Steps

Build replication environment – configure one master and two slaves.

Configure SSH password‑less login – on each host run ssh-keygen and then

ssh-copy-id -i /root/.ssh/id_rsa -p 22 [email protected]

ssh-copy-id -i /root/.ssh/id_rsa -p 22 [email protected]

ssh-copy-id -i /root/.ssh/id_rsa -p 22 [email protected]

.

Install MHA‑node – on every server execute

yum -y install perl-DBD-MySQL ncftp perl-DBI.x86

rpm -ivh mha4mysql-node-0.57-0.el7.noarch.rpm

.

Install MHA‑manager – on the designated manager host run

yum install epel-release-7-5.noarch

yum install perl-DBD-MySQL perl-Config-Tiny perl-Log-Dispatch perl-Parallel-ForkManager rrdtool perl-rrdtool rrdtool-devel perl-Params-Validate

rpm -ivh mha4mysql-manager-0.57-0.el7.noarch.rpm

. (CentOS‑specific; adjust for other distros.)

Create MHA configuration file – edit /etc/mha/mha.cnf with sections such as [server default] (user, password, work directories, ssh_user, repl_user, etc.) and individual server definitions, marking candidates with candidate_master=1 or no_master=1.

Validate SSH – run masterha_check_ssh --conf=/etc/mha/mha.cnf.

Validate replication – run masterha_check_repl --conf=/etc/mha/mha.cnf.

. If the error "Can't exec \"mysqlbinlog\"" appears, create symbolic links on all servers:

ln -s /usr/local/mysql/bin/mysqlbinlog /usr/local/bin/mysqlbinlog

ln -s /usr/local/mysql/bin/mysql /usr/local/bin/mysql

(adjust paths for your environment).

Start manager – nohup masterha_manager --conf=/etc/mha/mha.cnf & and check status with masterha_check_status --conf=/etc/mha/mha.cnf.

Failover test – stop MySQL on the original master (192.168.31.204), then examine the manager log ( /home/mha/manager.log) and the slave status on 192.168.31.101. The new master IP should be 192.168.31.168, confirming that slave1 was promoted and replication threads are running.

Download links for MHA packages and documentation are provided at the end of the article.