Re: [hobbit] RE: strange rrd results via ncv

["Ralph Mitchell" <ralphmitchell (at) gmail.com>]

On Jan 21, 2008 4:09 AM, Whilding, Craig <
Craig_Whilding_TERM (at) relay1.mentorg.com> wrote:

>  I've had to reply to myself as hobbit mail isn't getting through to me
> for some reason.
>
>
>
> Gary or others do you know how I can get it to chart the exact vale that
> comes in then. It's a bit silly having a graph of 4.8 jobs etc. Also 4.5is the average of 4 and 5 so it doesn't make much sense anyway, Also it's
> the data at the bottom of the graph that is really odd – queue max = 1.5m,
> running average 445.0m?
>
>
>
> Not sure where that data is coming from?
>
The problem is that rrdtool expects data to arrive at specific intervals.
Any data that doesn't arrive at the precise time it should, gets
interpolated - i.e. "adjusted to fit the curve".  So, if the rrd is built
with a start time of midnight and update intervals of 300 seconds (5
minutes), you'd need to be delivering updates at 00:05, 00:10, 00:15, etc,
in order to get the exact values stored.

You'll see this in some of the standard graphs too.  I have a machine that's
recording network traffic and right now its current bits-per-second value on
the graph is 15247.9.  0.9 of a bit is ridiculous, but there it is...

Ralph Mitchell