MHA

This Vagrant setup allows you to test easily MHA [1]. MHA's wiki is already full of explanations on how to use it.

However, this is a small summary.

Centos 6.3's Vagrant Box

To get ready, the easiest to do is to download the Centos63 box I've used if you don't have any yet:

wget http://dl.dropbox.com/u/11697684/centos63.box

And add it:

vagrant box add centos63 centos63.box

Start the boxes

Because I didn't get time yet to find why grants are not added at the first provisioning, this is the easiest way to start the boxes and have everything reading for MHA:

vagrant up percona1
vagrant provision percona1
vagrant up percona2
vagrant up percona3
vagrant provision percona2
vagrant provision percona3

Then you should have percona1 as master and the other two as slaves (percona2 and percona3)

MHA needs an already running replication setup.

Some MHA commands to test

Check [2] is the SSH communication between the machines is correct:

masterha_check_ssh --conf=/etc/app1.cnf

Start MHA:

masterha_manager --conf=/etc/app1.cnf

Then you can kill the master (killall -9 mysqld_safe mysqld) and see mha in action.

Check the status:

masterha_check_status --conf=/etc/app1.cnf

example:

[root@percona1 ~]# masterha_check_status --conf=/etc/app1.cnf
app1 (pid:3969) is running(0:PING_OK), master:percona1

OR

[root@percona1 ~]# masterha_check_status --conf=/etc/app1.cnf
app1 is stopped(2:NOT_RUNNING).

Stop MHA:

masterha_stop --conf=/etc/app1.cnf

Switch master to a slave while the master is still alive:

masterha_master_switch --master_state=alive --conf=/etc/app1.cnf --new_master_host=percona3

Switch master to a slave when the master is dead (mysql must not respond on it):

masterha_master_switch --master_state=dead --interactive=0 --wait_on_failover_error=0 --dead_master_host=percona3 --new_master_host=percona1

[1]	http://code.google.com/p/mysql-master-ha/

[2]	This needs to be run on percona1, it acts also as manager node