Cacti 0.8.6 eating diskspace

Wernicke · Post by **Wernicke** » Fri Dec 10, 2004 5:42 am

Here's some more statistic:

change db 2,299 60.38 0.06
delete 1,716,433 45,081.30 45.64 %
insert 1,720,617 45,191.19 45.75 %
replace 1,834 48.17 0.05 %
select 94,730 2,488.04 2.52 %
update 61,803 1,623.23 1.64 %

Post by **TheWitness** » Fri Dec 10, 2004 7:09 am

The reason that Cacti is generating so much more SQL traffic in 0.8.6 is because we are using the database to hold polling results prior to updating the RRD files. Then these entries are removed from the table.

Therefore, for every Data Source, you have 1-meny Insert(s) and 1-many Delete(s). Quite a bit of SQL activity comparted to prior results.

So that you know, this was done to prepare for a multiple-poller architecture to be released in 0.8.7 that will enable the Database to be on ServerA, the Web Server to be on ServerB and pollers to be on ServersA-n.

TheWitness

Mr.Burns · Post by **Mr.Burns** » Fri Dec 10, 2004 8:40 am

I had the same problem with mysql.log, but the logrotate was not doing his job correctly. Then I checked the logrotate config and now it is working fine.

rosch · Post by **rosch** » Fri Dec 10, 2004 8:46 am

Mr.Burns wrote:I had the same problem with mysql.log, but the logrotate was not doing his job correctly. Then I checked the logrotate config and now it is working fine.

I have the mysql.log also in lograte, but at present it did not rotate it has at present a size of 4KB. There is nearly no output there (for me), just start and stop messages from starting/stopping the database daemon.

Roland

Mr.Burns · Post by **Mr.Burns** » Fri Dec 10, 2004 9:10 am

Roland, take a look if is there any config in your logrotate.conf or logrotate.d that is not used any more (like squid, postgresql, or services that does not run in your cacti machine). Also you can force your logrotate and watch the verbose messages with this command:

logrotate -fv /etc/logrotate.conf

Cheers

Cacti 0.8.6 eating diskspace

Who is online