Database crash

[clahti]

Hello all:

I am currently on 1.3.3 (I am working on the 1.3.3 -> 1.3.4 RPM's and upgrade scripts). Unfortuately my several of mysql database tables got corrupted and although I was able to repair them successfully, I am concerned about this. I have several questions regarding our current state:

• our history table contains 4.3 million records and our trends table contains 3.2 million. Does this sound right for about 2 months we have been running zabbix?
• our items table contains 3800 records, this is really only about half the hosts and items we will eventually be monitoring, so is this is a large install and if so are there really any optimizations I should be considering?
• I have changed globally all of our update intervals to 120 or greater (up from 60) and history from 180 days down to 90 days. Should I see a reduction in the number of history records due to this? I am not sure what the housekeeper is supposed to do after a global change such as this
• I am running mysql 5.x so I have the option to convert the tables to innodb, would this be better?

[clahti]

correction...our history table has 428358815 records...this can't be right. That is 428 MILLION records after 2 months... ???

[Alexei]

Yes, you will see reduction in a growth of history and also stabilization of history size after 90 days.

[clahti]

Yes, you will see reduction in a growth of history and also stabilization of history size after 90 days.

So this is when the records move from the history table to the trend tables? How does this work? Are we still able to get graph data after the 90 days and if so why not cut down the stored history to say 30 days? What are the pro's and con's?