Nagios Grapher assumes that nagios "interval_length" is 60 (1 minute)

Nagios provides a configuration variable in it's main configuration file called "interval_length". This setting is like a master dial that controls the length of the other time variables throughout the Nagios configuration. As the Nagios manual says:

This is the number of seconds per "unit interval" used for timing in the scheduling queue, re-notifications, etc. "Units intervals" are used in the object configuration file to determine how often to run a service check, how often to re-notify a contact, etc.

Important: The default value for this is set to 60, which means that a "unit value" of 1 in the object configuration file will mean 60 seconds (1 minute). I have not really tested other values for this variable, so proceed at your own risk if you decide to do so!

 http://nagios.sourceforge.net/docs/3_0/configmain.html

Although the nagios author suggests that values other than 60 have not been tested, it is in fact very common to have other values there. It's particularly common to set this value to 1, and adjust the other time intervals throughout the Nagios system accordingly. This allows monitoring in granularity of up to 1 second if necessary.

Anyway, regardless of how you feel about that, the NagiosGrapher package makes the assumption that that interval_length parameter will remain at 60, meaning that it calculates the wrong step size when creating a rrd database for the parameter to be graphed. Because of this assumption the graph will move in step sizes of (interval_length * 60) - a parameter that is monitored once per minute therefore, may only have a graph resolution of once per hour.

The attached patch fixes NagiosGrapher's minute obsession by looking up the interval_length from Nagios' config file and using that as the basis for calculations.