Graph Scale and Consolidation Numbers Off

[ed.whitesell]

I created a few new templates to get data from some Cisco gear, and I'm having the same problem on both sets of data. The graph scale and the numbers reported in the GPRINT section are corect, but they are shown in "millions" when they should just be in "hundreds".

i.e. the attached graph shows 441.86m for `Peer in Mesages` when it should just be 441.86

I spent some time digging through the forums and the documentation, but couldn't find anything specific to this type of thing. I thought it might have something to do with the `Unit Exponent Value`, but that appears to only affect the y-axis.

Thanks.

[rony]

Check out: http://forums.cacti.net/viewtopic.php?t=8168

[ed.whitesell]

Ok, so it's saying "milli" instead of "millions", my bad there. Thanks for the link.

But my question still stands. Why is rrdtool converting them? Or, more importantly, how do I get it to stop? I know the values should be plain whole numbers, no units, based on when I manually poll the OIDs.

[gandalf]

Is it possible that you're talking about a GAUGE value instead of a COUNTER one? GAUGEs represent absolute values, e.g. the number of messages seen in the last 300 sec. Whereas COUNTER would be something like number of messages since reboot (that is: this number always increases).
And rrdtool "normalizes" the data to a second. That is, if you got 6 messages in 300 sec this is equal to 0,02 messages in a second (or: 20 m messages in a second).
You may create a CDEF like "CURRENT_DATA_SOURCE,300,* to have it again normalized to "a 5 min interval" or a CDEF like "CURRENT_DATA_SOURCE,60,*" to have it normalized to a minute.
Reinhard

[rony]

If you not like it to normalize the values, you can use the CDEF "Make Every 5 Minutes" to show you the value that was taken at polling. This is the action I would suggest.

[ed.whitesell]

Is it possible that you're talking about a GAUGE value instead of a COUNTER one?

No, they're all counters. In this specific instance: Total Incoming BGP Messages, Total Outgoing BGP Messages, Total Incoming BGP Updates and Total Outgoing BGP Updates. Per peer.

If you not like it to normalize the values, you can use the CDEF "Make Every 5 Minutes" to show you the value that was taken at polling. This is the action I would suggest.

I set this on the Graph Item Templates for this Graph Template. It looks to have cleared up the issue. I'll let it run for another 24-48 hours and see how it does. I guess I just assumed it would start out "dumb" (i.e. just reporting the integer returned from SNMP), and that I would not have to do any CDEF functions. But we all know what assumption is...

Thanks rony.

[nileshsawant]

can share BGP templates

Thanking You,
Nilesh