Why GAUGE of int values is a float?

art79 · Post by **art79** » Sun Aug 23, 2009 2:49 pm

Hi there

I thought that if my input method script produces only integers and my data template describes field as GAUGE it would be always integer on the graph.

Am i wrong? i thought that GAUGE is not averaged in any way over time, or is it?

- My script does cast to integer so im sure values are only ints.
- My data template describes fields as ints as well.
- I gather every 60 seconds and failure threshold is set to 300 seconds.

And my graph looks like this attached image.

What am i doing wrong?

THANKS A MILLION!!!!

BSOD2600 · Post by **BSOD2600** » Sun Aug 23, 2009 10:24 pm

All looks good except the keep alive data item. Please post a graph debug and data source debug outputs for that. Also an 'rrdtool info filename.rrd' output.

engeishi · Post by **engeishi** » Mon Aug 24, 2009 2:33 am

This happens because of specification of RRDtool.

In many case, The data collection is not performed according to the interval defined .
For example, there are a lot of devices to poll, it will take some time the script returns the result, and so on.
The data collection interval tends to become longer than the defined interval.

RRDtool corrects data to become a defined interval when the graph drawing.
Refer to the attached figure.

P.S. Sorry for my broken English. English is not my native language.

art79 · Post by **art79** » Mon Aug 24, 2009 3:14 pm

Wow! thats really cool answer.

That would explain everything :- )

Thanks a million!

Post by **gandalf** » Wed Aug 26, 2009 3:01 pm

engeishi wrote:This happens because of specification of RRDtool.

In many case, The data collection is not performed according to the interval defined .
For example, there are a lot of devices to poll, it will take some time the script returns the result, and so on.
The data collection interval tends to become longer than the defined interval.

RRDtool corrects data to become a defined interval when the graph drawing.
Refer to the attached figure.

P.S. Sorry for my broken English. English is not my native language.

That is indeed not only correct but well explained! I'd like to put this into cacti documentation; is this ok for you?
Reinhard

engeishi · Post by **engeishi** » Thu Aug 27, 2009 1:02 am

Of course, Yes!
it is very happy for me to contribute to the community.

Post by **gandalf** » Thu Aug 27, 2009 1:27 pm

Please see the result at http://docs.cacti.net/manual:087:8_rrdt ... n_integers
If you see improvements of the text, please let me know
Reinhard

engeishi · Post by **engeishi** » Thu Aug 27, 2009 7:55 pm

Awesome work!
Good example makes easy to understand the concept.
Thank you.

schurzi · Post by **schurzi** » Fri Aug 28, 2009 7:55 am

I had the same problem with integers which were displayed als float values. I used a CDEF to round them to integers again.

The CDEF is:
CURRENT_DATA_SOURCE,CURRENT_DATA_SOURCE,FLOOR,-,0.5,-,CURRENT_DATA_SOURCE,CURRENT_DATA_SOURCE,FLOOR,-,0.5,-,ABS,+,CURRENT_DATA_SOURCE,CEIL,CURRENT_DATA_SOURCE,FLOOR,IF

The long equation is because RRDTool does not know a round function. So I had to extract the fraction part of the value and then use FLOOR and CEIL to round either up or down.

From looking at my Graphs this would solve the visualization problems of this RRDTool "feature".

mgupta · Post by **mgupta** » Wed Nov 18, 2009 4:49 pm

[quote="schurzi"]I had the same problem with integers which were displayed als float values. I used a CDEF to round them to integers again.

The CDEF is:
CURRENT_DATA_SOURCE,CURRENT_DATA_SOURCE,FLOOR,-,0.5,-,CURRENT_DATA_SOURCE,CURRENT_DATA_SOURCE,FLOOR,-,0.5,-,ABS,+,CURRENT_DATA_SOURCE,CEIL,CURRENT_DATA_SOURCE,FLOOR,IF

The long equation is because RRDTool does not know a round function. So I had to extract the fraction part of the value and then use FLOOR and CEIL to round either up or down.

From looking at my Graphs this would solve the visualization problems of this RRDTool "feature".[/quote]

This is a very dumb question. ABS mentioned above is Absolute Value correct? I don't see a pre-defined CDEF for it.

schurzi · Post by **schurzi** » Thu Nov 19, 2009 9:20 am

mgupta wrote:This is a very dumb question. ABS mentioned above is Absolute Value correct? I don't see a pre-defined CDEF for it.

Exactly, ABS is for Absolute Vaule. It is a standard RRDTool function.

Maybe to explain the CDEF:

For example, RRDTool generates the Value 3.6:

CURRENT_DATA_SOURCE,CURRENT_DATA_SOURCE,FLOOR,-,
0.5,-,

3.6 - 3
0.6 - 0.5
=
0.1

CURRENT_DATA_SOURCE, CURRENT_DATA_SOURCE,FLOOR,-,
0.5,-,

3.6 - 3
0.6 - 0.5
=
0.1

ABS,+

0.1 + ABS(0.1)
=
0.2

CURRENT_DATA_SOURCE,CEIL

3.6 = 4

CURRENT_DATA_SOURCE,FLOOR

3.6 = 3

IF

The if looks at 0.2, if it is non 0 the Value from CEIL is returned, if it is 0 the Value from Floor is returned.

For 3.4 it looks that way:

3.4 - 3
0.4 - 0.5
= -0.1

-0.1 + ABS(-0.1)
= 0

...

mgupta · Post by **mgupta** » Thu Nov 19, 2009 2:16 pm

Thanks!

smlick · Post by **smlick** » Thu May 31, 2012 10:02 am

I understand why the system produce a non integer value but I didn't understand how to solve it.
For example I made a graph with SNMP generic OID for Asterisk to graph current calls but sometimes the value is not integer, how to solve?
Can you explain with simple words

or with some screenshot?

Regards
Alessio

Post by **gandalf** » Fri Jun 22, 2012 3:49 pm

You can't "solve" this.
See e.g. 1st link of my sig, RRDTool chapter for more help in case the picture from above is not enough.
But you can cut off fractional part by using a GPRINT preset
R.

smlick · Post by **smlick** » Sat Jun 23, 2012 1:11 am

But shurzi did it!!!! Some post before, but I didn't understand well how to set CDEF.

schurzi wrote:I had the same problem with integers which were displayed als float values. I used a CDEF to round them to integers again.

The CDEF is:
CURRENT_DATA_SOURCE,CURRENT_DATA_SOURCE,FLOOR,-,0.5,-,CURRENT_DATA_SOURCE,CURRENT_DATA_SOURCE,FLOOR,-,0.5,-,ABS,+,CURRENT_DATA_SOURCE,CEIL,CURRENT_DATA_SOURCE,FLOOR,IF
.

Regards
Alessio

Cacti

Why GAUGE of int values is a float?

Why GAUGE of int values is a float?

THANKS

Re: Why GAUGE of int values is a float?

Re: Why GAUGE of int values is a float?

Re: Why GAUGE of int values is a float?

Who is online