Hello,
I would like to get my database replication check technique peer reviewed. Any suggestions for improvement? It’s fairly simple, in truth.
We have had instances where the mysql master and slave both report in-sync, but data on the master is not consistent with data on the slave. So to me that spells one thing: Read the data, perform a checksum, compare.
So without further ado, here is what I do:
Code:
LOG=/var/log/POOH { mysqlrplsync --master=repli_chk:PASSWD@localhost:3306 --slaves=repli_chk:PASSWD@db-replicant:3306 | tee -a $LOG | grep -qF " [DIFF] "; } && { echo “ERROR in mysql replication.” 1>&2; exit 1; }
I created user “repli_chk” in accordance with the requirements at the bottom of this page: [URL=“http://dev.mysql.com/doc/mysql-utilities/1.5/en/mysqlrplsync.html”]http://dev.mysql.com/doc/mysql-utili...qlrplsync.html[/URL]
I run this job twice a week, on Sundays in the afternoon (the day before work starts), and at Friday at 7pm (a couple of hours after work ends and just before backups start).
Tags: None