mysqlcheck -ao & Zabbix DB's

[elvar]

Hey all, for those of you using MySQL & InnoDB as your backend storage systems I was wondering if you guys ever run 'mysqlcheck -ao' to optimize the DB and if so, how do you get around the false positive alerts & queue backlog? If I set up a maintenance period that includes data collection then my DB is too busy for the 'mysqlcheck -ao' to complete. If I don't collect data I end up with 3 hour+ gaps in my data collection. If I chose not to use a maintenance period at all the DB can't keep up with Zabbix due to the mysqlcheck happening and I receive tons of false positive alerts.

Is there a way to use 'mysqlcheck -ao' successfully on a production DB? I realize if this was clustered I could just run it on the non-active DB but that isn't the case right now


Kind regards,

[g.vecchi]

Same problem here.
I think the only solution to not stop DB is a Multi-Master configuration.

There could be another way to partially solve the problem:

- Master-slave configuration
- perform mysqlcheck optimization on Slave
- stop Master and drop databases
- dumpall Slave
- import Slave dump on Master

You lost events during importation, but you'll have not false positive alerts.