Doing no consolidation on data

[kerk]

Hi all,

I've created an RRA with step=1

RRA_Y5M.png (9.52 KiB) Viewed 1004 times

Subsequently, data template is created based on this RRA and use to graph a network interface.

After collecting the data for sometime, max reading of the graph shows 237.38M,

Y5M.png (31.94 KiB) Viewed 1004 times

However when I zoomed into that peak (red arrow), the max reading becomes 332.92M

Y5M_zoom.png (40.37 KiB) Viewed 1004 times

Since step=1 (no consolidation), I was expecting the max reading of the zoomed peak to be 237.38M as well. Is my understanding correct?

Hope someone can enlighten me.

Thank you

/kerk

[gandalf]

You should not use the AVERAGE consolidation function in case you are interested in MAX. So use MAX instead (or both, which I do prefer)
R.

[kerk]

Hi Reinhard,

I'm terribly sorry for the very late reply, didn't get a mail from the system of new message.

With step=1, I understand that the stored datapoint in the RRAs will be the same whether CF:AVERAGE or CF:MAX is applied on it.

Meaning, I have these datapoints at every 5 mins interval.
3 5 6 2 1
Since step=1,
the RRA with CF:AVERAGE applied on it will have 3 5 6 2 1 and
the RRA with CF:MAX applied on it will have 3 5 6 2 1 as well?

Hence shouldn't the MAX of CF:MAX and MAX of CF:AVERAGE of the printed datapoints be the same when graphed over a time period?

(Green touching the black lines?)

Max_n_avg.png (25.23 KiB) Viewed 950 times

I've read this posted. I thought I understand, but probably not yet
http://forums.cacti.net/viewtopic.php?f=8&t=35868

/kerk

[gandalf]

The post should indeed match your issue.
In case you simply want to do it _correct_:
edit the graph template
chnage the line/area items to use MAX instead of AVERAGE. The results will come "close", then ...
R.

[kerk]

Hi Reinhard,

I figured that there are 2 data consolidation that can take place.

1st is when the RRA is defined. Here I use step=1, hence no consolidation.

2nd is when the chart is drawn. Data points might be consolidated depending on the time frame (horizontal px) of the displayed chart. For my yearly chart, from the downloaded csv, it shows that step=3600 (1hr). When I zoomed in until day-view, all the 5 mins interval datapoints can fit into the chart, that is when values of MAX of AVERAGE and MAX match, black line touching the top of the green area.

Hopefully my understanding is right this time.

/kerk

[gandalf]

2nd is when the chart is drawn. Data points might be consolidated depending on the time frame (horizontal px) of the displayed chart.

Gotcha! That's what I call "graphical consolidation". By default, the method for graphical consolidation is taken from the DEF in use (e.g. AVERAGE, even if you _want_ MAX). RRDTool allows for overriding, but current Cacti does not support that. This will come in near future ...

For my yearly chart, from the downloaded csv, it shows that step=3600 (1hr). When I zoomed in until day-view, all the 5 mins interval datapoints can fit into the chart, that is when values of MAX of AVERAGE and MAX match, black line touching the top of the green area.

Yep. for e.g. a daily graph, MAX(AVERAGE) = MAX(MAX) as long as no consolidation happens ...

So we're back at using a DEF with the MAX consolidation ...
R.

[kerk]

Sir,

Thank you for the confirmation.

/kerk