[rrd-users] unexpected NANs

[Eduardo M. Bragatto]

Edward Quick wrote:

> I'm having problems with some of my rrd's because all the PDP's are nans 
> eventhough the values which go into each PDP are present at the required 
> 60 second intervals. A few values may come in slightly over the 60 
> seconds, but I have set the heartbeat 120, so as long as there is data 
> within this time, I would have expected to get a PDP. Can anyone see 
> where I'm going wrong?
(..)
> rra[0].cf = "AVERAGE"
> rra[0].rows = 26352
> rra[0].pdp_per_row = 5
> rra[0].xff = 5.0000000000e-01
> rra[0].cdp_prep[0].value = NaN
> rra[0].cdp_prep[0].unknown_datapoints = 2
> rra[0].cdp_prep[1].value = NaN
> rra[0].cdp_prep[1].unknown_datapoints = 2

I can see here is that you don't have a RRA with a single step (to store 
each 1 min value). Do you really want not to have 1min average graphs? 
In that case, the finest data will be 5 steps average (5min by your 
definition).

> head -30 NT_MEMORY##NT_MEMORY##MEMmmPgsInptPrSc.csv
> 1195985611,0.016667
> 1195985671,2.01667
> 1195985738,7.08955
> 1195985799,0
> 1195985859,0
> 1195985919,0.737705
> 1195985982,0
> 1195986043,0
> 1195986103,0
> 1195986163,0
> 1195986227,0

If you're using those values the way they are you will have the NaN 
problems because rrdtool uses only integers for input. Maybe in your 
case -- that all values seem to be low -- it would be better to multiple 
them by 10 a few times instead of rounding them.

[]s
Eduardo M. Bragatto.