Recommended MySQL statistics for production environment

Other than usual system stats (i/o, ram usage, cpu, load, etc) I am currently collecting questions, qps, running queries and buffer pool hit.

I find qps pretty useless since our production servers uptime is very high, and its average value.

I was wondering about best practices for statistics collection for mysql production server. What other stats should I collect/monitor to understand the load on my server and act quickly without putting more stress on it?

Edit :

I am not looking for a 3rd party solution, I am already using zabbix (and ability to create hand written scripts) to collect stats / monitor our mysql cluster. There is a list of possible stats for collection in this link. And of course there are stats which are not listed here and can be collected via shell scripts. The real question is what stats must be collected to monitor our cluster efficiently without creating an unnecessary junk full of statistics.

Example : should we get Qcache_hit / Qcache_hit + queries ratio to see if our tables are hot enough ?

I'd highly recommend using MONyog.


MONyog MySQL Monitor and Advisor is a "MySQL DBA in a box'' that helps
MySQL DBAs manage more MySQL servers, tune their current MySQL servers
and find and fix problems with their MySQL database applications
before they can become serious problems or costly outages.

Our DevOps team uses this extensively for both production and development. These guys have most of the "best practices" baked into the application, so we really don't need to get our hands too dirty in the DBA stuff.

StackExchange Database Administrators Q#19791, answer score: 3