how to NOT graph UNKNOWN DATA ???

[sneakza]

Hi,

i was wondering , sometimes some of my servers, dns, mail, NAT whatever the case may be loose connection to my cacti server. Which leads to gaps in graphs .. and quetions having to being ansewered at the end of the month meetings why there was gap in the graph, hence loss of service somewhere in the network and none not always able to remember what happened and when for all these occurrences...

Is there a way I could not graph any UNKNOWN DATA within cacti?

[BSOD2600]

Nope, nor can you properly 'fill' in those gaps.

[sneakza]

Nope, nor can you properly 'fill' in those gaps.

Are you sure about that???






gandalf
Developer


Joined: 02 Dec 2004
Posts: 9325
Location: Muenster, Germany
Posted: Fri Feb 22, 2008 6:20 am Post subject:

--------------------------------------------------------------------------------

Sorry to correct you, BSOD260, but you CAN.
The parameters are HEARTBEAT (max time allowed before rrdtool uses NaN for missing values) and xff (max amount of missing values allowed).
But in each case, I do NOT recommend using them. After some time, you won't know what is REAL data and what is FAKED data
Reinhard

[peecee]

Hi,

i was wondering , sometimes some of my servers, dns, mail, NAT whatever the case may be loose connection to my cacti server. Which leads to gaps in graphs .. and quetions having to being ansewered at the end of the month meetings why there was gap in the graph, hence loss of service somewhere in the network and none not always able to remember what happened and when for all these occurrences...

Is there a way I could not graph any UNKNOWN DATA within cacti?

What would you propose you do with the gaps? Tell management that the time period did not occur? Seriously, if you have gaps in the graphs due to service outage or interruption, it would seem that you need a ticketing system to track the problems. Cacti just reports the data made available to it. Don't fake the numbers, because it will lead to confusion about what it really is displaying.

[sneakza]

Id prefer to have perfect looking graphs in this instance.. even though we dont leave in a perfect world

so if I could convert all UNKNOWN DATA to zeros which is possible using a CDEF calculation, would it be possible to convert that UNKNOWN DATA to say the avarage value of all the data and then graph that ?

and im not looking to hide hours worth of gaps just when a device looses 6 to 7 polls or how ever much to create a noticable gap. I know cacti wont be able to determine the difference between the lenghs in gaps but thats the price to pay for gapless graphs...

[gandalf]

Please see http://www.vandenbogaerdt.nl for some hints
Reinhard

[sneakza]

Please see http://www.vandenbogaerdt.nl for some hints
Reinhard

Code: Select all

If you do like the behaviour described in the previous chapter, where unknown intervals are computed as if they had the average rate, you need to alter unknowns into average rates. Here's how to do that:

VDEF:ds0average=ds0,AVERAGE 
CDEF:ds0modified=ds0,UN,ds0average,ds0,IF 
VDEF:ds0total=ds0modified,TOTAL 
GPRINT:ds0total:Total\:%lf 

This computes the average rate, then uses this to cover up unknowns. The modified array is then used to compute the total amount of data.

This is perfect exactly what I was looking for ...

does cacti support VDEF?, would like to keep my graphs all in cacti if possible..

[sneakza]

Just did some reading and by the looks of it VDEF support in 0.88 maybe .. am i right ?

Thanks for your help Reinhard

[gandalf]

Just did some reading and by the looks of it VDEF support in 0.88 maybe .. am i right ?

Yep. If you want the replacement by only a "0", you should have found the method in that link as well
Reinhard