MHA switching (recommended operation process)

Preface

If used mha structure mysql colony , For various reasons, you may encounter the need to migrate

So let's share that mha Switch The process of

Tip: current version MHA Manager 0.56

Summary

Checklist

Be sure to check the database status before switching , Here is a checklist ：

1. All node databases must be running properly

2. All standby database replications must point to the primary database , And the replication status is normal （io , sql Process status ok ）

3. The standby database is in replication status Seconds_Behind_Master Parameter cannot be greater than 3 Second

4. In the current situation keepalived Work well ,master get VIP, The candidate master On keepalived The operating state of is normal , And priority ratio master low

5. Database relay_log_purge The parameters are off , Otherwise, there will be potential loss relay log The risk of

6. Except the main library , On other standby databases read_only The parameters are on , Otherwise, there is a risk that the primary and standby databases are inconsistent

Manual switching operation manual

1. Turn off the background switch monitoring masterha_stop --conf=/etc/app1.cnf

2. Check the running status of the database

3. Check database parameters

4. Check keepalived Working state ,ip Mounting situation

5. Record to be cut slave( The candidate master) Of binlogfile and position( For subsequent synchronization )

6. Conduct repl check

7. Implement switching

8. Status check (mysql,keepalived, Parameters , sync )

9. system maintenance , Database maintenance

10. Primary backup synchronization ( The same as before slave position)

11. Add the maintained database to mha colony

12. Check the running status of the database

13. Check database parameters

14. modify keepalived.conf file , Lower the priority , Make it different from the current master master Scramble for ip

15. start-up keepalived, Check ip, Check keepalived Running state

16. Record to be cut slave( The candidate master, primary master) Of binlogfile and position( For subsequent synchronization )

17. Conduct repl check

18. Implement switching

19. recovery mha framework ( Refer to the previous steps )

20. Back office mha monitor

21. Monitor and observe

Command summary

masterha_check_ssh --conf=/etc/app1.cnf

masterha_check_status --conf=/etc/app1.cnf

masterha_stop --conf=/etc/app1.cnf

masterha_check_repl --conf=/etc/app1.cnf

nohup masterha_manager --conf=/etc/app1.cnf --ignore_last_failover &

masterha_master_switch --master_state=alive --conf=/etc/app1.cnf --new_master_host=m1 --interactive=0

purge_relay_logs --user=root --password=xxxx --workdir=/data/relay_tmp/

show slave status\G

show master status;

show variables like 'relay_log_purge';

show variables like 'read_only';

Original address