Nagios bandwidth weathermap and perf data issue

[kyle.parker]

I have a service on my devices that polls the device and calculates BW usage in both % for the weather maps and actual values for graphing (ie. Mbps, Gbps, etc). It works great and my weather maps look awesome as the values will auto format, meaning it will say Mbps upto 999Mbps and as soon as it rolls over to 1000Mbps it will format it to say 1Gbps instead. The problem comes with the graphing in nagios. It seems that the weathermap and the graph both use the same perf data and therefore the graphs not only graph BW, but also the percentage value. I could live with this, I however can't live with the fact that the graphs don't auto scale. They seem to pick one scale and any value on that graph uses that scale, so if the values were both Gbps to start with than the scale will be messed up when it switches to Mbps or vice versa. So in my case I am monitoring 10Gige connection to our core network and during the night when everybody is sleeping the download is roughly 700Mbps and the upload is 400Mbps. Then when people wake up and start to use our network the download goes up to around 1.26Gbps and the upload doesn't change much. So on the graph it looks like the download suddenly drops to 1.26Mbps at 8am and jumps back up to 700Mbps at 1am the next night. I am looking for a way to rectify this issue without using two seperate services as I don't want to double my snmp traffic due to the number of devices we have on our network (probably close to 1500). Maybe there is a graphing tool add-on that is much better than the integrated one that you guys could recommend or there is a setting that let's the graphs auto scale better.

[abrist]

RRDs do not support being fed different metric measurements. I bet the issue is not just in your graphs, but in the RRDs themselves. What are you using to check these? I know MRTG has some methods of compensation for this.

[kyle.parker]

After speaking with a colleague we determined the easiest and fastest way would be to add a second service. One for graphing and one for bw usage to use in the weather maps. They both access the same rrd file, but just output different perf data. He assured me the extra few snmp gets wouldn't cause much more traffic. I am sure there is a prettier way to do it, but I have already spent too much time writing custom plugins and config wizards to spend any more time on it. Thanks for the quick reply though!

To answer your question though as to how we are implementing this, I have completely trashed MRTG (as it was causing major issues with the type of cards we were scanning, Calix) and wrote all custom wizards and plugins to gather the information into an rrd file located on the server. The plugin I used was check_iftraffic5.pl from the exchange and modified it quite a bit to suit our needs. The wizard I created from the ground up with much, much, MUCH trial and error (I crashed our server a couple of times, php loops are scary!).

[Box293]

If this is PNP doing the graphing, you could create a custom template which will display separate graphs for xbps and %.

If this is what you are after let me know and I'll point you to some helpful resources.

[kyle.parker]

I am just using the default graphing in Nagios XI 2014R2.7

[tmcdonald]

Are the separate services working for you? Trying to stick them both into one graph will definitely have some scaling issues as you mentioned, but splitting them should work fine.