Problems with graphs, help me.

[kurosaki]

Hello friends, sorry for the inconvenience. I still understand how cacti works. I am currently working with version 1.2.23 and I would like to see if it is possible to reset or clean the data from the graphs since they are currently wrong. I have attached an image of the ping latency of my server



and another image of how cacti graphs the result



I am currently using the unix ping template, hopefully someone could explain to me what is happening there in advance, thank you very much.

[TheWitness]

Well, the latency really depends on the number of packets/bytes that you are sending to the far end and expecting in return. If the number of north of 64 bytes (I think that's the default), the latency goes up.

You are slicing the banana into too many pieces. But it's important that you understand the difference. So, kudos to you for asking.

There is another layer too BTW. It's the RRDtool storage engine. You will eventually have more questions. We are here awaiting them.

[kurosaki]

Hello thank you very much for answering sorry if I can not understand your answer really do not know where to make changes to get the result I want or if this behavior that is generating cacti is normal, greetings.

[cigamit]

The graph looks correct for the latency you posted. The little m means milliseconds (which I know is confusing because the label is also milliseconds). RRDTool doesn't like to show tiny decimals on the y axis so it swaps to a small unit instead so it can make the numbers larger.

You can see the list of units here.
viewtopic.php?t=8168

For the data at the bottom, you can probably swap the graph to use the CDEF "Exact Numbers". For the Y Axis on the side, you would have to change the graph template for the "--units-exponent" option to set it to 0.