avoid spikes from being removed from graphs

[ghattasjacob]

Helolo Guys,
I don't know if I am even phrasing it right...
I have an issue, which seems a feature I want to avoid.
I have lots of interfaces which have legit spikes which I want to display in the graphs.
I see the spikes in the "daily"graph but when it moves to the weekly, monthly or yearly, it seems like cacti is averaging the data.
how can I avoid it or make the "spikes" appear more?
10xs And Best Regards.
Jacob

[bmfmancini]

What you are seeing is the averaging of the data like you said
to fix this you will have to modify the averaging of the data profile to store more detailed info

by default the daily 1 minute graph will store that detail for 2 days and then consolodate to 15 minute averages so you will lose momentary spikes

Check out the doc here
https://github.com/Cacti/documentation/ ... rofiles.md

This will make your graphs bigger so plan ahead !

[cigamit]

There is consolidation happening on multiple levels. First, as Sean points out, as data moves from one Profile RRA to the next, is it averaged into the new time slots.

The 2nd time actually happens via RRDTool visually. So even if you extend out the per minute data storage to 1 month, If you attempt to view 1 month of data at once, you have way more data points than RRDTool can display on the screen. So for instance say you only had 1800 data points, yet the RRDTool graph is 300 pixels wide. RRDTool can only fit 1 data point per pixel, so it is going to average your 1800 data points to fit within the available pixels (300). The data is still there of course if you Zoom in, since by Zooming in, you are reducing the number of data points that need to be displayed.

Now if you would rather things weren't averaged down, you can get into the complex world of "MAX of AVERAGES", etc... by changing up some things. Whether that is possible ofcourse depends on your Cacti Version (which you never state what version you are using).

[TheWitness]

You have to be careful with spikekill. The most important part is choosing the right method and limits. Before setting up profilactic despiking, you really need to experiment with settings.