Cacti and per minute poll

I love the tool Cacti. It’s a nice graphing tool, which helps a lot in system monitoring and management.

Its logic, however, is not always obvious. The relationship between data types, host templates and graph templates is quite complex, and if you are to implement any non-default setup using Cacti, you better understand these relationships as soon as possible.

I was using it to monitor an ERP system. I have encountered in the past a problem with monitoring Linux network traffic through SNMP, since the container is a short integer, and during 5 minutes wait it gets filled several times, if this is 1Gb/s network connection we measure.

The solution I wanted to employ was to haste the system, and measure every minute (60 seconds). This isn’t trivial using either Cacti or MRTG, because they were designed for 5 minute interval.

I have found this link in Cacti’s forums, which shows how to do it, and it works fine, as long as you follow instructions by the letter. This was done on RH4 x86, using Dag Wieers’ Home-made RPM repository as the source for Cacti.

I will attempt to export the modified templates and upload them here, to save some of the dirty work required for the process to work.