Combining Traffic and Errors on a Graph for Cisco Switches

[Dog]

Hi

I am trying to combine Traffic and Error Graphs for Cisco Switches (2950 series) to appear on the one graph, this will allow me to look at one graph per interface as opposed to two.

Running Linux 7.3 and Cacti-0.8.1

Thus far I have tried

1) Create graph template, basically copy default errors/discards template
and add traffic info as specified in the default traffic template.

2)Create data query - Use the existing XML path as indicated in SNMP interface querries - Add the Graph that I created in step 1 - Select Data Sources Discards In, Discards Out, Errors In, Errors Out, Traffic In, Traffic Out - Tick boxes and add names and titles in the suggested values as indicated in the SNMP interface queries for traffic an Discards / Errors.

3) Create a host template select the data query created in step 2.

When the crontab runs I receive a segmentation error - Data is collected though it does not appear to be written to the MYSQL database as the files in the rra directory are not updated.

Am I going about this the right way ?

On another issue does anyone know where the graphs are ordered - ie
the snmp index does not order 1 to X - changes can be made using the graph trees though this can be time consuming when doing 48port switches.

Thanx Dog.

[Dog]

OK OK so no one wants to talk to the dog (growl is larger than the bite). For other reservior dogs I was able to graph Errors and Traffic on the one graph by the following.

One of the pitfalls is the errors are expressed in mill and micro so they do not appear on the graph when plotting against the meg data traffic.

Create a CEF to multiply by 10 000 000 -> Copy the CEF * 1024, replace the 1024 with 10 000 000.

Setup a polling host use the default Cisco Router -> Poll the device twice on the first pass poll for traffic, on the second pass poll for errors/discards.

Create a graph template -> Duplicate the errors/discards graph template
and rename to blah. Edit the blah template set the error and discard values to use the CEF that you created above and add traffic details as per the standard interface traffic template.

Under graph management select the host that you are polling and add the
graph blah. Select all the interface details save the graph. Add the new graph to the appropriate tree and your away.

I would hazzard a guess that there is a better way to do this.

Cheers

[bulek]

There is no better way I guess. You are just trying to plot completly different data series on one graph. What do you have in Y-axis description: "bits per sec" or "errors per sec"? The only correct way would be to use second Y-axis but it is not available in rrdtool.

You may want to multiply it by -10mln instead... this way you will have error statistics on negative part of the graph and they will be separated from traffic stats a bit.

- bulek